diff --git a/.pre-commit-config.yaml b/.pre-commit-config.yaml
index 72ddaded..c9a1723d 100644
--- a/.pre-commit-config.yaml
+++ b/.pre-commit-config.yaml
@@ -72,7 +72,6 @@ repos:
           - fastapi>=0.109 # Match pyproject.toml
           - fhaviary>=0.8.2 # Match pyproject.toml
           - httpx
-          - litellm>=1.40.15,!=1.49.4,!=1.49.5,!=1.49.6 # Match pyproject.toml
           - numpy>=1.20 # Match pyproject.toml
           - pydantic~=2.0 # Match pyproject.toml
           - tenacity
diff --git a/ldp/alg/optimizer/ape.py b/ldp/alg/optimizer/ape.py
index b0eceaad..110eb7b4 100644
--- a/ldp/alg/optimizer/ape.py
+++ b/ldp/alg/optimizer/ape.py
@@ -7,6 +7,8 @@
 from typing import Any, Self, cast
 
 from aviary.core import Message
+from llmclient import LLMResult
+from llmclient import MultipleCompletionLLMModel as LLMModel
 from pydantic import (
     BaseModel,
     ConfigDict,
@@ -21,7 +23,6 @@
 from ldp.alg.optimizer.opt import Optimizer
 from ldp.data_structures import Trajectory
 from ldp.graph import LLMCallOp, OpResult, PromptOp
-from ldp.llms import LLMModel, LLMResult
 
 logger = logging.getLogger(__name__)
 
@@ -285,7 +286,7 @@ async def _get_updated_prompt(
                 )
             ),
         ]
-        result = await self.llm.call(messages, output_type=OutputPrompt)
+        result = await self.llm.call_single(messages, output_type=OutputPrompt)
         message_content = cast(str, cast(list[Message], result.messages)[-1].content)
         try:
             return OutputPrompt.model_validate_json(message_content).prompt
diff --git a/ldp/graph/common_ops.py b/ldp/graph/common_ops.py
index 25b4ac6d..ac83b2e0 100644
--- a/ldp/graph/common_ops.py
+++ b/ldp/graph/common_ops.py
@@ -13,16 +13,15 @@
 import numpy as np
 import tree
 from aviary.core import Message, Tool, ToolRequestMessage, is_coroutine_callable
-from pydantic import BaseModel
-
-from ldp.llms import (
+from llmclient import (
     EmbeddingModel,
     HybridEmbeddingModel,
-    LiteEmbeddingModel,
-    LLMModel,
+    LiteLLMEmbeddingModel,
     LLMResult,
     SparseEmbeddingModel,
 )
+from llmclient import MultipleCompletionLLMModel as LLMModel
+from pydantic import BaseModel
 
 from .gradient_estimators import assign_constant_grads
 from .memory import Memory, MemoryModel, UIndexMemoryModel
@@ -254,7 +253,9 @@ async def forward(
             # if no tools are provided, tool_choice must be 'none'
             tool_choice = "none"
 
-        result = await model.call(messages=msgs, tools=tools, tool_choice=tool_choice)
+        result = await model.call_single(
+            messages=msgs, tools=tools, tool_choice=tool_choice
+        )
         if result.messages is None:
             raise ValueError("No messages returned")
 
@@ -306,7 +307,7 @@ async def compute_logprob(
         # are consistent - not sure we'd be sampling from the same distribution as N independent samples.
         # TODO: think about whether sampling params besides temperature need to be accounted for, like top_p
         results = await asyncio.gather(*[
-            model.call(temperature=1, **model_kwargs)
+            model.call_single(temperature=1, **model_kwargs)
             for _ in range(self.num_samples_partition_estimate)
         ])
         temp_factor = 1.0 / temperature - 1.0
@@ -406,18 +407,18 @@ def __init__(
         emb_models: list[EmbeddingModel] = []
         if dense_embedding_dim > 0:
             emb_models.append(
-                LiteEmbeddingModel(
+                LiteLLMEmbeddingModel(
                     name=dense_embedding,
-                    dimensions=dense_embedding_dim,
+                    ndim=dense_embedding_dim,
                     embed_kwargs=embedding_model_kwargs,
                 )
             )
         if sparse_embedding_dim > 0:
-            emb_models.append(SparseEmbeddingModel(dimensions=sparse_embedding_dim))
+            emb_models.append(SparseEmbeddingModel(ndim=sparse_embedding_dim))
         self.embedding = HybridEmbeddingModel(models=emb_models)
 
     async def forward(self, string_input: str) -> np.ndarray:
-        return await self.embedding.embed_text(string_input)
+        return np.array(await self.embedding.embed_document(string_input))
 
     @classmethod
     def backward(
diff --git a/ldp/graph/memory.py b/ldp/graph/memory.py
index 5e820576..293f66cb 100644
--- a/ldp/graph/memory.py
+++ b/ldp/graph/memory.py
@@ -7,6 +7,7 @@
 
 import numpy as np
 import numpy.typing as npt
+from llmclient import EmbeddingModel
 from pydantic import (
     BaseModel,
     ConfigDict,
@@ -18,8 +19,6 @@
 )
 from usearch.index import Index
 
-from ldp.llms import EmbeddingModel
-
 if TYPE_CHECKING:
     from .common_ops import MemoryOp
     from .op_utils import CallID
@@ -128,7 +127,7 @@ def enforce_empty(cls, v: dict) -> dict:
 
     async def add_memory(self, memory: Memory) -> None:
         key = await self._add_to_index(
-            embedding=await self.embedding_model.embed_text(memory.query)
+            embedding=np.array(await self.embedding_model.embed_document(memory.query))
         )
         self.memories[key] = memory
 
@@ -138,7 +137,8 @@ async def get_memory(
         self, query: str, matches: int = DEFAULT_MEMORY_MATCHES
     ) -> list[Memory]:
         return await self._search_index(
-            embedding=await self.embedding_model.embed_text(query), matches=matches
+            embedding=np.array(await self.embedding_model.embed_document(query)),
+            matches=matches,
         )
 
     def __len__(self) -> int:
@@ -148,7 +148,7 @@ def __len__(self) -> int:
     async def safe_access_index(self) -> AsyncIterator[TIndex]:
         """Get the internal Index under the protection of an internal Lock."""
         # pylint bug, SEE: https://github.com/pylint-dev/pylint/issues/9813
-        async with self._index_lock:  # pylint: disable=not-async-context-manager
+        async with self._index_lock:
             yield self._index
 
     @abstractmethod
@@ -167,9 +167,9 @@ class UIndexMemoryModel(MemoryModel[Index]):
 
     def __init__(self, **kwargs):
         super().__init__(**kwargs)
-        if not self.embedding_model.dimensions:
+        if not self.embedding_model.ndim:
             raise TypeError("Specify dimensions to the embedding model.")
-        self._index = Index(ndim=self.embedding_model.dimensions)
+        self._index = Index(ndim=self.embedding_model.ndim)
 
     async def _add_to_index(self, embedding: np.ndarray) -> int:
         async with self.safe_access_index() as index:
diff --git a/ldp/graph/modules/reflect.py b/ldp/graph/modules/reflect.py
index 40af0bd7..ab60a603 100644
--- a/ldp/graph/modules/reflect.py
+++ b/ldp/graph/modules/reflect.py
@@ -5,8 +5,7 @@
 
 from ldp.graph import ConfigOp, FxnOp, LLMCallOp, PromptOp, compute_graph
 from ldp.graph.ops import ResultOrValue
-from ldp.llms import append_to_sys
-from ldp.llms.prompts import indent_xml
+from ldp.llms import append_to_sys, indent_xml
 
 
 class ReflectModuleConfig(BaseModel):
diff --git a/ldp/llms/__init__.py b/ldp/llms/__init__.py
index 7795d172..d4df5bb6 100644
--- a/ldp/llms/__init__.py
+++ b/ldp/llms/__init__.py
@@ -1,21 +1,24 @@
-from .chat import (
-    JSONSchemaValidationError,
-    LLMModel,
+from llmclient import (
     LLMResult,
-    MultipleCompletionLLMModel,
     sum_logprobs,
     validate_json_completion,
 )
-from .embeddings import (
+from llmclient import MultipleCompletionLLMModel as LLMModel
+from llmclient.embeddings import (
     EmbeddingModel,
     EmbeddingModes,
     HybridEmbeddingModel,
-    LiteEmbeddingModel,
+    LiteLLMEmbeddingModel,
     SparseEmbeddingModel,
 )
+from llmclient.exceptions import (
+    JSONSchemaValidationError,
+)
+
 from .prompts import (
     append_to_messages,
     append_to_sys,
+    indent_xml,
     prepend_sys,
     prepend_sys_and_append_sys,
 )
@@ -27,11 +30,11 @@
     "JSONSchemaValidationError",
     "LLMModel",
     "LLMResult",
-    "LiteEmbeddingModel",
-    "MultipleCompletionLLMModel",
+    "LiteLLMEmbeddingModel",
     "SparseEmbeddingModel",
     "append_to_messages",
     "append_to_sys",
+    "indent_xml",
     "prepend_sys",
     "prepend_sys_and_append_sys",
     "sum_logprobs",
diff --git a/ldp/llms/chat.py b/ldp/llms/chat.py
deleted file mode 100644
index 4aa506a5..00000000
--- a/ldp/llms/chat.py
+++ /dev/null
@@ -1,371 +0,0 @@
-import asyncio
-import json
-from collections.abc import AsyncGenerator, Callable, Iterable, Mapping
-from datetime import datetime
-from typing import Any, ClassVar, Self, TypeAlias, cast
-from uuid import UUID, uuid4
-
-import litellm
-from aviary.core import (
-    Message,
-    Tool,
-    ToolRequestMessage,
-    ToolsAdapter,
-    is_coroutine_callable,
-)
-from pydantic import BaseModel, ConfigDict, Field, ValidationError, model_validator
-
-# Yes, this is a hack, it mostly matches
-# https://github.com/python-jsonschema/referencing/blob/v0.35.1/referencing/jsonschema.py#L20-L21
-JSONSchema: TypeAlias = Mapping[str, Any]
-
-
-class JSONSchemaValidationError(ValueError):
-    """Raised when the completion does not match the specified schema."""
-
-
-class LLMResult(BaseModel):
-    """A class to hold the result of a LLM completion."""
-
-    id: UUID = Field(default_factory=uuid4)
-    config: dict | None = None
-    prompt: list[Message] | None = Field(
-        default=None, description="Messages sent to the LLM."
-    )
-    messages: list[Message] | None = Field(
-        default=None, description="Messages received from the LLM."
-    )
-    prompt_count: int = Field(default=0, description="Count of prompt tokens.")
-    completion_count: int = Field(default=0, description="Count of completion tokens.")
-    model: str
-    date: str = Field(default_factory=datetime.now().isoformat)
-    seconds_to_first_token: float | None = None
-    seconds_to_last_token: float = 0
-    logprob: float | None = Field(
-        default=None, description="Sum of logprobs in the completion."
-    )
-    system_fingerprint: str | None = Field(
-        default=None, description="System fingerprint received from the LLM."
-    )
-
-    @property
-    def prompt_and_completion_costs(self) -> tuple[float, float]:
-        """Get a two-tuple of prompt tokens cost and completion tokens cost, in USD."""
-        return litellm.cost_per_token(
-            self.model,
-            prompt_tokens=self.prompt_count,
-            completion_tokens=self.completion_count,
-        )
-
-    @property
-    def provider(self) -> str:
-        """Get the model provider's name (e.g. "openai", "mistral")."""
-        return litellm.get_llm_provider(self.model)[1]
-
-    def get_supported_openai_params(self) -> list[str] | None:
-        """Get the supported OpenAI parameters for the model."""
-        return litellm.get_supported_openai_params(self.model)
-
-
-def sum_logprobs(choice: litellm.utils.Choices) -> float | None:
-    """Calculate the sum of the log probabilities of an LLM completion (a Choices object).
-
-    Args:
-        choice: A sequence of choices from the completion.
-
-    Returns:
-        The sum of the log probabilities of the choice.
-    """
-    try:
-        logprob_obj = choice.logprobs
-    except AttributeError:
-        return None
-    if isinstance(logprob_obj, dict):
-        if logprob_obj.get("content"):
-            return sum(
-                logprob_info["logprob"] for logprob_info in logprob_obj["content"]
-            )
-    elif choice.logprobs.content:
-        return sum(logprob_info.logprob for logprob_info in choice.logprobs.content)
-    return None
-
-
-def validate_json_completion(
-    completion: litellm.ModelResponse, output_type: type[BaseModel] | JSONSchema
-) -> None:
-    """Validate a completion against a JSON schema.
-
-    Args:
-        completion: The completion to validate.
-        output_type: A JSON schema or a Pydantic model to validate the completion.
-    """
-    try:
-        for choice in completion.choices:
-            if not hasattr(choice, "message") or not choice.message.content:
-                continue
-            # make sure it is a JSON completion, even if None
-            # We do want to modify the underlying message
-            # so that users of it can just parse it as expected
-            choice.message.content = (
-                choice.message.content.split("```json")[-1].split("```")[0] or ""
-            )
-            if isinstance(output_type, Mapping):  # JSON schema
-                litellm.litellm_core_utils.json_validation_rule.validate_schema(
-                    schema=dict(output_type), response=choice.message.content
-                )
-            else:
-                output_type.model_validate_json(choice.message.content)
-    except ValidationError as err:
-        raise JSONSchemaValidationError(
-            "The completion does not match the specified schema."
-        ) from err
-
-
-class MultipleCompletionLLMModel(BaseModel):
-    """Run n completions at once, all starting from the same messages."""
-
-    model_config = ConfigDict(extra="forbid")
-
-    # this should keep the original model
-    # if fine-tuned, this should still refer to the base model
-    name: str = "unknown"
-    config: dict = Field(
-        default={
-            "model": "gpt-3.5-turbo",  # Default model should have cheap input/output for testing
-            "temperature": 0.1,
-        }
-    )
-    encoding: Any | None = None
-
-    def __str__(self) -> str:
-        return f"{type(self).__name__} {self.name}"
-
-    @model_validator(mode="after")
-    def set_model_name(self) -> Self:
-        if (
-            self.config.get("model") in {"gpt-3.5-turbo", None}
-            and self.name != "unknown"
-        ) or (self.name != "unknown" and "model" not in self.config):
-            self.config["model"] = self.name
-        elif "model" in self.config and self.name == "unknown":
-            self.name = self.config["model"]
-        # note we do not consider case where both are set
-        # because that could be true if the model is fine-tuned
-        return self
-
-    async def achat(
-        self, messages: Iterable[Message], **kwargs
-    ) -> litellm.ModelResponse:
-        return await litellm.acompletion(
-            messages=[m.model_dump(by_alias=True) for m in messages],
-            **(self.config | kwargs),
-        )
-
-    async def achat_iter(self, messages: Iterable[Message], **kwargs) -> AsyncGenerator:
-        return cast(
-            AsyncGenerator,
-            await litellm.acompletion(
-                messages=[m.model_dump(by_alias=True) for m in messages],
-                stream=True,
-                stream_options={
-                    "include_usage": True,  # Included to get prompt token counts
-                },
-                **(self.config | kwargs),
-            ),
-        )
-
-    # SEE: https://platform.openai.com/docs/api-reference/chat/create#chat-create-tool_choice
-    # > `none` means the model will not call any tool and instead generates a message.
-    # > `auto` means the model can pick between generating a message or calling one or more tools.
-    # > `required` means the model must call one or more tools.
-    NO_TOOL_CHOICE: ClassVar[str] = "none"
-    MODEL_CHOOSES_TOOL: ClassVar[str] = "auto"
-    TOOL_CHOICE_REQUIRED: ClassVar[str] = "required"
-    # None means we won't provide a tool_choice to the LLM API
-    UNSPECIFIED_TOOL_CHOICE: ClassVar[None] = None
-
-    async def call(  # noqa: C901, PLR0915
-        self,
-        messages: list[Message],
-        callbacks: list[Callable] | None = None,
-        output_type: type[BaseModel] | JSONSchema | None = None,
-        tools: list[Tool] | None = None,
-        tool_choice: Tool | str | None = TOOL_CHOICE_REQUIRED,
-        **chat_kwargs,
-    ) -> list[LLMResult]:
-        start_clock = asyncio.get_running_loop().time()
-
-        # Deal with tools. Note OpenAI throws a 400 response if tools is empty:
-        # > Invalid 'tools': empty array. Expected an array with minimum length 1,
-        # > but got an empty array instead.
-        # So, circumvent this behavior if tools in (None, [])
-        if tools:
-            chat_kwargs["tools"] = ToolsAdapter.dump_python(
-                tools, exclude_none=True, by_alias=True
-            )
-            if tool_choice is not None:
-                chat_kwargs["tool_choice"] = (
-                    {
-                        "type": "function",
-                        "function": {"name": tool_choice.info.name},
-                    }
-                    if isinstance(tool_choice, Tool)
-                    else tool_choice
-                )
-
-        if isinstance(output_type, Mapping):  # Use structured outputs
-            chat_kwargs["response_format"] = {
-                "type": "json_schema",
-                "json_schema": {
-                    "strict": True,
-                    # SEE: https://platform.openai.com/docs/guides/structured-outputs#additionalproperties-false-must-always-be-set-in-objects
-                    "schema": dict(output_type) | {"additionalProperties": False},
-                    "name": output_type["title"],  # Required by OpenAI as of 12/3/2024
-                },
-            }
-        elif output_type is not None:  # Use JSON mode
-            schema = json.dumps(output_type.model_json_schema(mode="serialization"))
-            schema_msg = f"Respond following this JSON schema:\n\n{schema}"
-            # Get the system prompt and its index, or the index to add it
-            i, system_prompt = next(
-                ((i, m) for i, m in enumerate(messages) if m.role == "system"),
-                (0, None),
-            )
-            messages = [
-                *messages[:i],
-                (
-                    system_prompt.append_text(schema_msg, inplace=False)
-                    if system_prompt
-                    else Message(role="system", content=schema_msg)
-                ),
-                *messages[i + 1 if system_prompt else i :],
-            ]
-            chat_kwargs["response_format"] = {"type": "json_object"}
-
-        # add static configuration to kwargs
-        chat_kwargs = self.config | chat_kwargs
-        n = chat_kwargs.get("n", 1)  # number of completions
-        if n < 1:
-            raise ValueError("Number of completions (n) must be >= 1.")
-
-        prompt = [
-            (
-                m
-                if not isinstance(m, ToolRequestMessage) or m.tool_calls
-                # OpenAI doesn't allow for empty tool_calls lists, so downcast empty
-                # ToolRequestMessage to Message here
-                else Message(role=m.role, content=m.content)
-            )
-            for m in messages
-        ]
-        results: list[LLMResult] = []
-
-        if callbacks is None:
-            completion: litellm.ModelResponse = await self.achat(prompt, **chat_kwargs)
-            if output_type is not None:
-                validate_json_completion(completion, output_type)
-
-            for choice in completion.choices:
-                if isinstance(choice, litellm.utils.StreamingChoices):
-                    raise NotImplementedError("Streaming is not yet supported.")
-
-                if (
-                    tools is not None  # Allows for empty tools list
-                    or choice.finish_reason == "tool_calls"
-                    or (getattr(choice.message, "tool_calls", None) is not None)
-                ):
-                    serialized_choice_message = choice.message.model_dump()
-                    serialized_choice_message["tool_calls"] = (
-                        serialized_choice_message.get("tool_calls") or []
-                    )
-                    output_messages: list[Message | ToolRequestMessage] = [
-                        ToolRequestMessage(**serialized_choice_message)
-                    ]
-                else:
-                    output_messages = [Message(**choice.message.model_dump())]
-
-                results.append(
-                    LLMResult(
-                        model=self.name,
-                        config=chat_kwargs,
-                        prompt=prompt,
-                        messages=output_messages,
-                        logprob=sum_logprobs(choice),
-                        system_fingerprint=completion.system_fingerprint,
-                        # Note that these counts are aggregated over all choices
-                        completion_count=completion.usage.completion_tokens,  # type: ignore[attr-defined,unused-ignore]
-                        prompt_count=completion.usage.prompt_tokens,  # type: ignore[attr-defined,unused-ignore]
-                    )
-                )
-        else:
-            if tools:
-                raise NotImplementedError("Using tools with callbacks is not supported")
-            if n > 1:
-                raise NotImplementedError(
-                    "Multiple completions with callbacks is not supported"
-                )
-            result = LLMResult(model=self.name, config=chat_kwargs, prompt=prompt)
-
-            sync_callbacks = [f for f in callbacks if not is_coroutine_callable(f)]
-            async_callbacks = [f for f in callbacks if is_coroutine_callable(f)]
-            stream_completion = await self.achat_iter(messages, **chat_kwargs)
-            text_result = []
-            role = "assistant"
-
-            async for chunk in stream_completion:
-                delta = chunk.choices[0].delta
-                role = delta.role or role
-                if delta.content:
-                    s = delta.content
-                    if result.seconds_to_first_token == 0:
-                        result.seconds_to_first_token = (
-                            asyncio.get_running_loop().time() - start_clock
-                        )
-                    text_result.append(s)
-                    [await f(s) for f in async_callbacks]
-                    [f(s) for f in sync_callbacks]
-                if hasattr(chunk, "usage"):
-                    result.prompt_count = chunk.usage.prompt_tokens
-
-            output = "".join(text_result)
-            result.completion_count = litellm.token_counter(
-                model=self.name,
-                text=output,
-            )
-            # TODO: figure out how tools stream, and log probs
-            result.messages = [Message(role=role, content=output)]
-            results.append(result)
-
-        if not results:
-            # This happens in unit tests. We should probably not keep this block around
-            # long-term. Previously, we would emit an empty ToolRequestMessage if
-            # completion.choices were empty, so  I am replicating that here.
-            results.append(
-                LLMResult(
-                    model=self.name,
-                    config=chat_kwargs,
-                    prompt=prompt,
-                    messages=[ToolRequestMessage(tool_calls=[])],
-                )
-            )
-
-        end_clock = asyncio.get_running_loop().time()
-
-        for result in results:
-            # Manually update prompt count if not set, which can
-            # happen if the target model doesn't support 'include_usage'
-            if not result.prompt_count:
-                result.prompt_count = litellm.token_counter(
-                    model=self.name,
-                    messages=[m.model_dump() for m in result.messages],  # type: ignore[union-attr]
-                )
-
-            # update with server-side counts
-            result.seconds_to_last_token = end_clock - start_clock
-
-        return results
-
-
-class LLMModel(MultipleCompletionLLMModel):
-    async def call(self, *args, **kwargs) -> LLMResult:  # type: ignore[override]
-        return (await super().call(*args, **kwargs))[0]
diff --git a/ldp/llms/embeddings.py b/ldp/llms/embeddings.py
deleted file mode 100644
index d809948d..00000000
--- a/ldp/llms/embeddings.py
+++ /dev/null
@@ -1,134 +0,0 @@
-import asyncio
-from abc import ABC, abstractmethod
-from enum import StrEnum
-from typing import Any
-
-import litellm
-import numpy as np
-import tiktoken
-from pydantic import BaseModel, ConfigDict, Field, model_validator
-
-
-class EmbeddingModes(StrEnum):
-    """Enum representing the different modes of an embedding model."""
-
-    DOCUMENT = "document"
-    QUERY = "query"
-
-
-class EmbeddingModel(ABC, BaseModel):
-    name: str
-    dimensions: int | None = None
-
-    def set_mode(self, mode: EmbeddingModes) -> None:
-        """Several embedding models have a 'mode' or prompt which affects output."""
-
-    @abstractmethod
-    async def embed_texts(self, texts: list[str]) -> list[np.ndarray]:
-        pass
-
-    async def embed_text(self, text: str) -> np.ndarray:
-        return (await self.embed_texts([text]))[0]
-
-    @staticmethod
-    def from_name(embedding: str, **kwargs) -> "EmbeddingModel":
-        if embedding.startswith("hybrid"):
-            dense_model = LiteEmbeddingModel(name="-".join(embedding.split("-")[1:]))
-            return HybridEmbeddingModel(
-                name=embedding, models=[dense_model, SparseEmbeddingModel(**kwargs)]
-            )
-        if embedding == "sparse":
-            return SparseEmbeddingModel(**kwargs)
-        return LiteEmbeddingModel(name=embedding, **kwargs)
-
-
-class LiteEmbeddingModel(EmbeddingModel):
-    name: str = Field(default="text-embedding-3-small")
-    dimensions: int | None = Field(
-        default=None,
-        description=(
-            "The length an embedding will have. If left unspecified, we attempt to"
-            " infer an un-truncated length via LiteLLM's internal model map. If this"
-            " inference fails, the embedding will be un-truncated."
-        ),
-    )
-    batch_size: int = 16
-    embed_kwargs: dict[str, Any] = Field(
-        default_factory=dict,
-        description="Extra kwargs to pass to litellm.aembedding.",
-    )
-
-    @model_validator(mode="before")
-    @classmethod
-    def infer_dimensions(cls, data: dict[str, Any]) -> dict[str, Any]:
-        if data.get("dimensions") is not None:
-            return data
-        # Let's infer the dimensions
-        config: dict[str, dict[str, Any]] = litellm.get_model_cost_map(
-            url="https://raw.githubusercontent.com/BerriAI/litellm/main/litellm/model_prices_and_context_window_backup.json"
-        )
-        output_vector_size: int | None = config.get(data.get("name", ""), {}).get(  # noqa: FURB184
-            "output_vector_size"
-        )
-        if output_vector_size:
-            data["dimensions"] = output_vector_size
-        return data
-
-    async def embed_texts(self, texts: list[str]) -> list[np.ndarray]:
-        embeddings = []
-        # Before you get excited to asyncio.gather this:
-        # The point of this is to not hit the API rate limit
-        for i in range(0, len(texts), self.batch_size):
-            response = await litellm.aembedding(
-                model=self.name,
-                input=texts[i : i + self.batch_size],
-                encoding_format="float",
-                dimensions=self.dimensions,
-                **self.embed_kwargs,
-            )
-            embeddings.extend([
-                np.array(e["embedding"], dtype=np.float32) for e in response.data
-            ])
-        return embeddings
-
-
-class SparseEmbeddingModel(EmbeddingModel):
-    """This is a very simple keyword search model - probably best to be mixed with others."""
-
-    model_config = ConfigDict(arbitrary_types_allowed=True)
-
-    name: str = "sparse"
-    dimensions: int = 256
-    enc: tiktoken.Encoding = Field(
-        default_factory=lambda: tiktoken.get_encoding("cl100k_base")
-    )
-
-    async def embed_texts(self, texts) -> list[np.ndarray]:
-        enc_batch = self.enc.encode_ordinary_batch(texts)
-        # now get frequency of each token rel to length
-        return [
-            np.bincount(
-                [xi % self.dimensions for xi in x], minlength=self.dimensions
-            ).astype(np.float32)
-            / len(x)
-            for x in enc_batch
-        ]
-
-
-class HybridEmbeddingModel(EmbeddingModel):
-    name: str = "hybrid-embed"
-    models: list[EmbeddingModel]
-
-    @model_validator(mode="before")
-    @classmethod
-    def infer_dimensions(cls, data: dict[str, Any]) -> dict[str, Any]:
-        if data.get("dimensions") is not None:
-            raise ValueError(f"Don't specify dimensions to {cls.__name__}.")
-        if not data.get("models") or any(m.dimensions is None for m in data["models"]):
-            return data
-        data["dimensions"] = sum(m.dimensions for m in data["models"])
-        return data
-
-    async def embed_texts(self, texts):
-        all_embeds = await asyncio.gather(*[m.embed_texts(texts) for m in self.models])
-        return np.concatenate(all_embeds, axis=1)
diff --git a/ldp/utils.py b/ldp/utils.py
index d00c761b..96eadd04 100644
--- a/ldp/utils.py
+++ b/ldp/utils.py
@@ -1,38 +1,7 @@
 import logging
 import logging.config
-import os
 from typing import Any
 
-import litellm
-
-
-def configure_log_levels() -> None:
-    """Configure log levels."""
-    # Set sane default LiteLLM logging configuration
-    # SEE: https://docs.litellm.ai/docs/observability/telemetry
-    litellm.telemetry = False
-    if (
-        logging.getLevelNamesMapping().get(
-            os.environ.get("LITELLM_LOG", ""), logging.WARNING
-        )
-        < logging.WARNING
-    ):
-        # If LITELLM_LOG is DEBUG or INFO, don't change the LiteLLM log levels
-        litellm_loggers_config: dict[str, Any] = {}
-    else:
-        litellm_loggers_config = {
-            "LiteLLM": {"level": "WARNING"},
-            "LiteLLM Proxy": {"level": "WARNING"},
-            "LiteLLM Router": {"level": "WARNING"},
-        }
-
-    logging.config.dictConfig({
-        "version": 1,
-        "disable_existing_loggers": False,
-        # Lower level for httpx and LiteLLM
-        "loggers": {"httpx": {"level": "WARNING"}} | litellm_loggers_config,
-    })
-
 
 def configure_stdout_logs(
     name: str = "root",
diff --git a/pyproject.toml b/pyproject.toml
index a046d47c..c3adf380 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -24,9 +24,9 @@ classifiers = [
 dependencies = [
     "aiofiles",
     "dm-tree",
+    "fh-llm-client",
     "fhaviary>=0.8.2",  # For core namespace
     "httpx",
-    "litellm>=1.52.15",  # For litellm.ssl_verify
     "networkx[default]~=3.4",  # Pin for pydot fix
     "numpy>=1.20",  # For numpy.typing
     "pydantic~=2.0",
@@ -48,7 +48,7 @@ dev = [
     "fhaviary[xml]",
     "ipython>=8",  # Pin to keep recent
     "ldp[monitor,nn,rich,server,typing,visualization]",
-    "litellm!=1.49.4,!=1.49.5,!=1.49.6",  # For https://github.com/BerriAI/litellm/issues/6216
+    "litellm>=1.52.15",
     "mypy>=1.8",  # Pin for mutable-override
     "pre-commit>=3.4",  # Pin to keep recent
     "pydantic~=2.9",  # Pydantic 2.9 changed JSON schema exports 'allOf', so ensure tests match
diff --git a/tests/cassettes/TestMemoryOpt.test_lessons_memory_optimizer.yaml b/tests/cassettes/TestMemoryOpt.test_lessons_memory_optimizer.yaml
index effff073..b8861fac 100644
--- a/tests/cassettes/TestMemoryOpt.test_lessons_memory_optimizer.yaml
+++ b/tests/cassettes/TestMemoryOpt.test_lessons_memory_optimizer.yaml
@@ -1,9 +1,8 @@
 interactions:
   - request:
       body:
-        '{"messages": [{"role": "user", "content": "Guess a number based on the
-        input word."}, {"role": "user", "content": "Previous attempts:\n\n-----\n\nHello"}],
-        "model": "gpt-4-turbo", "temperature": 0}'
+        '{"messages":[{"role":"user","content":"Guess a number based on the input
+        word."},{"role":"user","content":"Previous attempts:\n\n-----\n\nHello"}],"model":"gpt-4-turbo","n":1,"temperature":0}'
       headers:
         accept:
           - application/json
@@ -12,13 +11,13 @@ interactions:
         connection:
           - keep-alive
         content-length:
-          - "197"
+          - "191"
         content-type:
           - application/json
         host:
           - api.openai.com
         user-agent:
-          - AsyncOpenAI/Python 1.54.3
+          - AsyncOpenAI/Python 1.56.2
         x-stainless-arch:
           - arm64
         x-stainless-async:
@@ -28,11 +27,11 @@ interactions:
         x-stainless-os:
           - MacOS
         x-stainless-package-version:
-          - 1.54.3
+          - 1.56.2
         x-stainless-raw-response:
           - "true"
         x-stainless-retry-count:
-          - "1"
+          - "0"
         x-stainless-runtime:
           - CPython
         x-stainless-runtime-version:
@@ -42,22 +41,22 @@ interactions:
     response:
       body:
         string: !!binary |
-          H4sIAAAAAAAAA4xTTY/TMBC991cMvgDatGrSpuxW6qGLkBYJBMtyWKCocuxJYnBsy56wVKv+d+Sm
-          bVo+JC62NG/emw8/Pw4AmJJsDkzUnETj9HD54U4a/rKavL3+9P6+kP6hVsXtrLiqb++nLIkMW3xD
-          QQfWSNjGaSRlTQcLj5wwqqYvJmk6vpyk+Q5orEQdaZWj4XRIrS/sMBtn0+F4Ohxf7dm1VQIDm8OX
-          AQDA4+6MfRqJP9kcxskh0mAIvEI2PyYBMG91jDAeggrEDbGkB4U1hGbX+g1qbZ/ARwtViyEAB9M2
-          BXooeEAJ1gDVCMq4luDBegmrjpKsGLx+qjW0AYFDUHF4aJBqK6HYQKxbGWUqQC5q0EiEvlNHrwTX
-          8IPrFvsyigI4G1RcIKiuLNeu5gUSPFsu0gSuF1kCo9Eogc+LbPYcuJExzUBomyaWohoDdsJhtDIr
-          cwMLuFyZV7CAfGXewALSrL/fxTuPeUspez7ZCqlGP4dLuIAcLiDNjkcepbLIubPJ+ZJ+X0+z2e+0
-          tL7fqwqQZ6PT1/BYtoFHR5hW6318e3xebSvnbRH2+DFeKqNCvfbIgzXxKQNZx3bodgDwdWej9swZ
-          zHnbOFqT/Y4mCmZ5J8d68/Zgmu09xsgS1yfA9EA701tLJK50ODEiE1zUKHtq71reSmVPgMHJ1H+2
-          8zftbnJlqv+R7wEh0BHKtfMolTgfuU/zGD/3v9KOW941zMImEDbrUpkKvfOq+1qlW+eymIwns0lZ
-          ssF28AsAAP//AwAIUB2raAQAAA==
+          H4sIAAAAAAAAAwAAAP//jFNdjxIxFH3nV1z7omYHMswyCCQ8YLKKyWY3cX1aMaTTXmaqnba2d1Tc
+          8N9NGWDAj8SXNrnnnnM/evrUA2BKshkwUXEStdP9hUzvy59vF4/5ZPx1dJu/CXKq7h7u3i+nNwVL
+          IsMWn1HQkTUQtnYaSVnTwsIjJ4yqw1fX15N8Op5ke6C2EnWklY76oz41vrD9LM1G/XTUT6cHdmWV
+          wMBm8LEHAPC0P2OfRuIPNoM0OUZqDIGXyGanJADmrY4RxkNQgbghlnSgsIbQ7Ftfotb2GXywUDYY
+          AnAwTV2gh4IHlGANUIWgjGsIvlsvYdVSkhWDd8+1hiYgcAgqDg81UmUlFFuIdUujTAnIRQUaidC3
+          6uiV4Bq+cd1gV0ZRAGeDigsE1Zbl2lW8QIIXi/kwgdfzLIHBYJDA4zwbvwRuZEwzEJq6jqWowoCt
+          cBiszMosYQ6TlbmBOeQrcwtzGGbdfR/vPOYtpOz4ZEukCv0MJnAFOVzBMDsdeZTKIufBJpdL+n09
+          9faw04313V5VgDwbnL+Gx00TeHSEabQ+xHen59W2dN4W4YCf4htlVKjWHnmwJj5lIOvYHt31AD7t
+          bdRcOIM5b2tHa7Jf0ETBLG/lWGfeDhxmB48xssT1GTA60i701hKJKx3OjMgEFxXKjtq5ljdS2TOg
+          dzb1n+38TbudXJnyf+Q7QAh0hHLtPEolLkfu0jzGz/2vtNOW9w2zsA2E9XqjTIneedV+rY1bp5tx
+          mmFRSM56u94vAAAA//8DAOGAL6hoBAAA
       headers:
         CF-Cache-Status:
           - DYNAMIC
         CF-RAY:
-          - 8df9523c6fee17e6-SJC
+          - 8effb64679542289-SJC
         Connection:
           - keep-alive
         Content-Encoding:
@@ -65,9 +64,15 @@ interactions:
         Content-Type:
           - application/json
         Date:
-          - Fri, 08 Nov 2024 23:25:19 GMT
+          - Tue, 10 Dec 2024 19:41:26 GMT
         Server:
           - cloudflare
+        Set-Cookie:
+          - __cf_bm=mMdeNhgsijzLQWo1eS7oWCKtWSlI7vGGbMfe387TW2o-1733859686-1.0.1.1-DUZOKAzl79eqH_N1Fv4.foYQ9VPZ.wGufm_V8WVD.HwzlWQkvcEJSAah_NbuiZEeEagW4JINDfDr3Y.MaKhp7g;
+            path=/; expires=Tue, 10-Dec-24 20:11:26 GMT; domain=.api.openai.com; HttpOnly;
+            Secure; SameSite=None
+          - _cfuvid=caE6vH.B1YK84P8vrxF1dX_gwfnPB9lipTGU3rCdq30-1733859686280-0.0.1.1-604800000;
+            path=/; domain=.api.openai.com; HttpOnly; Secure; SameSite=None
         Transfer-Encoding:
           - chunked
         X-Content-Type-Options:
@@ -79,33 +84,32 @@ interactions:
         openai-organization:
           - future-house-xr4tdh
         openai-processing-ms:
-          - "3379"
+          - "3867"
         openai-version:
           - "2020-10-01"
         strict-transport-security:
           - max-age=31536000; includeSubDomains; preload
         x-ratelimit-limit-requests:
-          - "3000"
+          - "10000"
         x-ratelimit-limit-tokens:
-          - "250000"
+          - "2000000"
         x-ratelimit-remaining-requests:
-          - "2999"
+          - "9999"
         x-ratelimit-remaining-tokens:
-          - "249964"
+          - "1999963"
         x-ratelimit-reset-requests:
-          - 20ms
+          - 6ms
         x-ratelimit-reset-tokens:
-          - 8ms
+          - 1ms
         x-request-id:
-          - req_6acb7ca6b0eedb5dd7042354724ae613
+          - req_cbc976d488c682233276db4ac2c209fd
       status:
         code: 200
         message: OK
   - request:
       body:
-        '{"messages": [{"role": "user", "content": "Guess a number based on the
-        input word."}, {"role": "user", "content": "Previous attempts:\n\n-----\n\nDay"}],
-        "model": "gpt-4-turbo", "temperature": 0}'
+        '{"messages":[{"role":"user","content":"Guess a number based on the input
+        word."},{"role":"user","content":"Previous attempts:\n\n-----\n\nDay"}],"model":"gpt-4-turbo","n":1,"temperature":0}'
       headers:
         accept:
           - application/json
@@ -114,13 +118,13 @@ interactions:
         connection:
           - keep-alive
         content-length:
-          - "195"
+          - "189"
         content-type:
           - application/json
         host:
           - api.openai.com
         user-agent:
-          - AsyncOpenAI/Python 1.54.3
+          - AsyncOpenAI/Python 1.56.2
         x-stainless-arch:
           - arm64
         x-stainless-async:
@@ -130,7 +134,7 @@ interactions:
         x-stainless-os:
           - MacOS
         x-stainless-package-version:
-          - 1.54.3
+          - 1.56.2
         x-stainless-raw-response:
           - "true"
         x-stainless-retry-count:
@@ -144,19 +148,19 @@ interactions:
     response:
       body:
         string: !!binary |
-          H4sIAAAAAAAAA4xSQW7bMBC86xULXnqxAlmRU9u3pkaKoocWza1JIdDkSmJKcQkuidQI/PdCsmPZ
-          aAr0wsPMznBmyZcMQBgt1iBUJ6Pqvc0/fL/XofWf2G/ubzfFt6fUfLn78dWmu2f+KGaDgrZPqOKr
-          6kpR7y1GQ+5Aq4Ay4uA6f389nxfL6/lqJHrSaAdZ62Ne5TGFLeVlUVZ5UeXF6qjuyChksYaHDADg
-          ZTyHnE7jb7GGYvaK9MgsWxTr0xCACGQHREhmw1G6KGYTqchFdGP0W8mogRzEDuGZgoZHsZG72aOA
-          z++shTYh80i61G8xQFnNQI5IQJABoaygoxQYjAMJWu6uzq8K2CSWQ12XrD3i+1N2S60PtOUjf8Ib
-          4wx3dUDJ5IacHMmLkd1nAD/HHaWL2sIH6n2sI/1CNxiWi4OdmF7mDTJSlHbCF8e9XrrVGqM0ls92
-          LJRUHepJOT2ITNrQGZGddf47zFveh97Gtf9jPxFKoY+oax9QG3VZeBoLOPzbf42ddjwGFrzjiH3d
-          GNdi8MEcfk3ja73Qy+pmubpBke2zPwAAAP//AwCpLLoXQwMAAA==
+          H4sIAAAAAAAAAwAAAP//jFKxbtswFNz1FQ9cukiBIstu4i1NOnRqiy4tmkKgyCeJCcXHkE9xjcD/
+          Xkh2LBtNgS4c7t4d7x75kgAIo8UahOokq97b7Ebnnx+fvupvt8rrxfc7/PHUq81z+bz5mH8R6aig
+          +gEVv6ouFPXeIhtye1oFlIyj6+X7xeJqeb26Wk1ETxrtKGs9Z2XGQ6gpK/KizPIyy68P6o6MwijW
+          8DMBAHiZzjGn0/hbrCFPX5EeY5QtivVxCEAEsiMiZIwmsnQs0plU5BjdFP2DjKiBHHCHsKGg4V7c
+          yW16L+DTO2uhHTDGiXRDX2OAokxBTkhAkAGhKKGjIUQwDiRoub04vSpgM0Q51nWDtQd8d8xuqfWB
+          6njgj3hjnIldFVBGcmPOyOTFxO4SgF/Tjoaz2sIH6j1XTI/oRsNiubcT88u8QTKxtDO+POz13K3S
+          yNLYeLJjoaTqUM/K+UHkoA2dEMlJ57/DvOW9721c+z/2M6EUekZd+YDaqPPC81jA8d/+a+y44ymw
+          iNvI2FeNcS0GH8z+1zS+yptVXmBdaymSXfIHAAD//wMAOWqgiUMDAAA=
       headers:
         CF-Cache-Status:
           - DYNAMIC
         CF-RAY:
-          - 8df95252cb1017e6-SJC
+          - 8effb65f7b522289-SJC
         Connection:
           - keep-alive
         Content-Encoding:
@@ -164,7 +168,7 @@ interactions:
         Content-Type:
           - application/json
         Date:
-          - Fri, 08 Nov 2024 23:25:20 GMT
+          - Tue, 10 Dec 2024 19:41:27 GMT
         Server:
           - cloudflare
         Transfer-Encoding:
@@ -178,33 +182,32 @@ interactions:
         openai-organization:
           - future-house-xr4tdh
         openai-processing-ms:
-          - "1385"
+          - "708"
         openai-version:
           - "2020-10-01"
         strict-transport-security:
           - max-age=31536000; includeSubDomains; preload
         x-ratelimit-limit-requests:
-          - "3000"
+          - "10000"
         x-ratelimit-limit-tokens:
-          - "250000"
+          - "2000000"
         x-ratelimit-remaining-requests:
-          - "2999"
+          - "9999"
         x-ratelimit-remaining-tokens:
-          - "249964"
+          - "1999963"
         x-ratelimit-reset-requests:
-          - 20ms
+          - 6ms
         x-ratelimit-reset-tokens:
-          - 8ms
+          - 1ms
         x-request-id:
-          - req_f3bdd2b659f522b3a80f9b62e2a8789a
+          - req_46e66130e4ac25fcd1f0215ce253d587
       status:
         code: 200
         message: OK
   - request:
       body:
-        '{"messages": [{"role": "user", "content": "Guess a number based on the
-        input word."}, {"role": "user", "content": "Previous attempts:\n\n-----\n\nBar"}],
-        "model": "gpt-4-turbo", "temperature": 0}'
+        '{"messages":[{"role":"user","content":"Guess a number based on the input
+        word."},{"role":"user","content":"Previous attempts:\n\n-----\n\nBar"}],"model":"gpt-4-turbo","n":1,"temperature":0}'
       headers:
         accept:
           - application/json
@@ -213,13 +216,13 @@ interactions:
         connection:
           - keep-alive
         content-length:
-          - "195"
+          - "189"
         content-type:
           - application/json
         host:
           - api.openai.com
         user-agent:
-          - AsyncOpenAI/Python 1.54.3
+          - AsyncOpenAI/Python 1.56.2
         x-stainless-arch:
           - arm64
         x-stainless-async:
@@ -229,7 +232,7 @@ interactions:
         x-stainless-os:
           - MacOS
         x-stainless-package-version:
-          - 1.54.3
+          - 1.56.2
         x-stainless-raw-response:
           - "true"
         x-stainless-retry-count:
@@ -243,19 +246,19 @@ interactions:
     response:
       body:
         string: !!binary |
-          H4sIAAAAAAAAA4xSsW7bMBDd9RUHLl2kQJbsJvGWdMpQoEiCLnUhUORJYkvxWPKEpA3874Vkx1La
-          BMjC4b17D+/d8SkBEEaLLQjVSVa9t9nV7Z2OV+H+5vHu62ep7J8vF/dcPPw6l5/oVqSjguofqPhZ
-          daao9xbZkDvQKqBkHF1X5+VqlV+URT4RPWm0o6z1nK0zHkJNWZEX6yxfZ/nlUd2RURjFFr4lAABP
-          0zvmdBofxRYmrwnpMUbZotiehgBEIDsiQsZoIkvHIp1JRY7RTdGvZUQN5IA7hAcKGnbiWoZ0J+Dm
-          g7XQDhjjRLqhrzFAmYKMYBh603YMAa1kBKZphruACBaZMUQw/7qe7cQyRsBmiHJchRusPeL7Uy9L
-          rQ9UxyN/whvjTOyqgDKSGztEJi8mdp8AfJ/2N7xYifCBes8V0090o2GxOdiJ+WoL8vJIMrG0M75Z
-          p6+4VRpZGhsX+xdKqg71rJyPJQdtaEEki87/h3nN+9DbuPY99jOhFHpGXfmA2qiXheexgOOffmvs
-          tOMpsIi/I2NfNca1GHwwhx/V+Gqj6zIvP5ZNI5J98hcAAP//AwAHHYUcXwMAAA==
+          H4sIAAAAAAAAA4xSy27bMBC86ysWvPQiBYpfsXyLURTtKYcCDYo6EChyJTGlSIK7aloE/vdCsmMp
+          bQr0wsPMzmBml88JgDBa7ECoVrLqgs1udX5n86cf3N8/VurzHdf3H79s923xYfP+q0gHha8eUfGL
+          6kr5Llhk492JVhEl4+B6fbNcbtfFZnszEp3XaAdZEzhbZdzHymeLfLHK8lWWF2d1641CEjv4lgAA
+          PI/vkNNp/Cl2kKcvSIdEskGxuwwBiOjtgAhJZIilY5FOpPKO0Y3R95JQg3fALcKTjxoOYi9jehDw
+          6Z210PRINJKu7yqMsExBEhiGzjQtQ0QrGYH9OMNtRASLzBgJzJ+uVwcxjxGx7kkOq3C9tWf8eOll
+          fROir+jMX/DaOENtGVGSd0MHYh/EyB4TgIdxf/2rlYgQfRe4ZP8d3WC4WJ/sxHS1GVmcSfYs7YSv
+          V+kbbqVGlsbSbP9CSdWinpTTsWSvjZ8Ryazz32He8j71Nq75H/uJUAoDoy5DRG3U68LTWMThT/9r
+          7LLjMbCgX8TYlbVxDcYQzelH1aHMsViqTb6RK5Eck98AAAD//wMApNW1P18DAAA=
       headers:
         CF-Cache-Status:
           - DYNAMIC
         CF-RAY:
-          - 8df9525c3edd17e6-SJC
+          - 8effb664d90d2289-SJC
         Connection:
           - keep-alive
         Content-Encoding:
@@ -263,7 +266,7 @@ interactions:
         Content-Type:
           - application/json
         Date:
-          - Fri, 08 Nov 2024 23:25:21 GMT
+          - Tue, 10 Dec 2024 19:41:29 GMT
         Server:
           - cloudflare
         Transfer-Encoding:
@@ -277,43 +280,42 @@ interactions:
         openai-organization:
           - future-house-xr4tdh
         openai-processing-ms:
-          - "816"
+          - "2204"
         openai-version:
           - "2020-10-01"
         strict-transport-security:
           - max-age=31536000; includeSubDomains; preload
         x-ratelimit-limit-requests:
-          - "3000"
+          - "10000"
         x-ratelimit-limit-tokens:
-          - "250000"
+          - "2000000"
         x-ratelimit-remaining-requests:
-          - "2999"
+          - "9999"
         x-ratelimit-remaining-tokens:
-          - "249964"
+          - "1999963"
         x-ratelimit-reset-requests:
-          - 20ms
+          - 6ms
         x-ratelimit-reset-tokens:
-          - 8ms
+          - 1ms
         x-request-id:
-          - req_626e2457e60ac7dfe254e4f3731892ff
+          - req_8ad74985c93735a2b9ad1a1c53c42b6f
       status:
         code: 200
         message: OK
   - request:
       body:
-        '{"messages": [{"role": "system", "content": "Respond following this JSON
-        schema:\n\n{\"description\": \"Entry for a lesson created from some example
-        data.\", \"properties\": {\"query\": {\"description\": \"Plain text string for
-        retrieving this lesson from a database of lessons.\", \"title\": \"Query\",
-        \"type\": \"string\"}, \"lesson\": {\"description\": \"Lesson generated from
-        past attempts.\", \"title\": \"Lesson\", \"type\": \"string\"}}, \"required\":
-        [\"query\", \"lesson\"], \"title\": \"LessonEntry\", \"type\": \"object\"}"},
-        {"role": "user", "content": "We are trying to guess a number based on the input
-        word. We just tried 3 times, and collected rewards where a higher reward is
-        better. Here are the results in three-tuples of input, output, reward:\n- (''Hello'',
-        ''Hello! To guess a number based on the input word \"Hello,\" I\\''ll use a
-        simple method by assigning each letter a numerical value based on its position
-        in the alphabet (A=1, B=2, ..., Z=26) and then summing these values.\\n\\nH
+        '{"messages":[{"role":"system","content":"Respond following this JSON schema:\n\n{\"description\":
+        \"Entry for a lesson created from some example data.\", \"properties\": {\"query\":
+        {\"description\": \"Plain text string for retrieving this lesson from a database
+        of lessons.\", \"title\": \"Query\", \"type\": \"string\"}, \"lesson\": {\"description\":
+        \"Lesson generated from past attempts.\", \"title\": \"Lesson\", \"type\": \"string\"}},
+        \"required\": [\"query\", \"lesson\"], \"title\": \"LessonEntry\", \"type\":
+        \"object\"}"},{"role":"user","content":"We are trying to guess a number based
+        on the input word. We just tried 3 times, and collected rewards where a higher
+        reward is better. Here are the results in three-tuples of input, output, reward:\n-
+        (''Hello'', ''Hello! To guess a number based on the input word \"Hello,\" I\\''ll
+        use a simple method by assigning each letter a numerical value based on its
+        position in the alphabet (A=1, B=2, ..., Z=26) and then summing these values.\\n\\nH
         = 8\\nE = 5\\nL = 12\\nL = 12\\nO = 15\\n\\nAdding these together: 8 + 5 + 12
         + 12 + 15 = 52\\n\\nSo, based on the word \"Hello,\" my guess for a number is
         52.'', -300.0)\n-(''Day'', ''Based on the word \"Day,\" I\\''ll guess the number
@@ -321,8 +323,7 @@ interactions:
         \"Bar,\" I\\''ll guess the number 3, as it might relate to the three letters
         in the word \"Bar.\"'', -100.0)\n\nPlease create a lesson based on this data
         referencing the relative success or failure associated with the reward, use
-        concise wording and don''t repeat yourself."}], "model": "gpt-4o-mini-2024-07-18",
-        "response_format": {"type": "json_object"}}'
+        concise wording and don''t repeat yourself."}],"model":"gpt-4o-mini-2024-07-18","n":1,"response_format":{"type":"json_object"}}'
       headers:
         accept:
           - application/json
@@ -331,13 +332,13 @@ interactions:
         connection:
           - keep-alive
         content-length:
-          - "1667"
+          - "1660"
         content-type:
           - application/json
         host:
           - api.openai.com
         user-agent:
-          - AsyncOpenAI/Python 1.54.3
+          - AsyncOpenAI/Python 1.56.2
         x-stainless-arch:
           - arm64
         x-stainless-async:
@@ -347,7 +348,7 @@ interactions:
         x-stainless-os:
           - MacOS
         x-stainless-package-version:
-          - 1.54.3
+          - 1.56.2
         x-stainless-raw-response:
           - "true"
         x-stainless-retry-count:
@@ -361,25 +362,25 @@ interactions:
     response:
       body:
         string: !!binary |
-          H4sIAAAAAAAAA4xUTY/jNgy951cQuszFGeRjvja3LoqiLbCX3S0waFMEtETb6tKSK1HJegfz3xey
-          nTiDtkAvPvCJj++RJl8WAMoatQOlGxTddrz84eMnI5+fVv69X/16+p2fP7XPv23Txw+nD/ysipzh
-          y79IyznrVvu2YxLr3QjrQCiUWdeP2/V69bTdrAeg9YY4p9WdLO/8srXOLjerzd1y9bhcP03Zjbea
-          otrBHwsAgJfhm3U6Q1/VDlbFOdJSjFiT2l0eAajgOUcUxmijoBNVzKD2TsgN0l/26u9Eod+rHexV
-          nShG62pwqS0pRCgxkgHvwLouCZx8MHGvCtgrphi9G9N+cSCNjTDGCjgRoEPuvxGgCLWdRBAPAzvg
-          RA5pqHRFfAufGwKfpEsS4YjBkingZKUBaQhKigKBYmKBknIuxui1zV2eX2UmuPmZmH1xA6fG6gZ6
-          S2zIAIJG1omHjFGNr+B+M9tkEqEAR+REsQBpfKobsLmuJnscONifINAJg8nZy+1qNQofK/+I/Q3U
-          5CgMVRCGZn+VhDyX3NwVEKiiQE5nI1n41BVfQeNTiGAdIBjsCyjTm/qRbd0I91COYq+0bLKWn6xD
-          5r64asd7DDfAZPIUcJaxvZqvxLN57ZOTYur0OKKBqrF1M47gUm89erdZrbEahSJIg5L7zgStDwSB
-          mI7o5NyICC3240hGyrODWAAyz39MFMs8ySCTVTiqUexx+Ee0b/OEqO0ajPbbuYu6QWZyNWV9bN2X
-          DAx/V/buUkvBauRpwkBVRTpTcn+7V6/XSxKoShHzorrEPMVfL1vHvu6CL+OEX+KVdTY2h0AYvcsb
-          FsV3akBfFwB/Dtud3iys6oJvOzmI/0IuE27fPY58aj4qM7p+2E6oeEGegfuH6Si8JTwYErQcrw6E
-          0qgbMnPqfE0wGeuvgMWV7X/K+Tfu0bp19f+hnwGtqRMyhy6Qsfqt5flZoHx0/+vZpc2DYBX7KNQe
-          KutqCl2w48mrusO78vGpfFjr+41avC6+AwAA//8DABQKeBEABgAA
+          H4sIAAAAAAAAAwAAAP//jFTBjttGDL37K4i5+CIbttfuet1T2iBo0aA5BEgOdWGMZ2iJyWioDik7
+          xmL/PRjJtrxoCvQiYPjI90iK5PMIwJA3GzCusurqJkze+NmH+PSruNUf7z/9Kd/efJwv3ra2cqvm
+          88kUOYL3X9DpNWrquG4CKnHsYZfQKmbW+ePDw3r19NP6qQNq9hhyWNnoZMmTmiJNFrPFcjJ7nMzX
+          l+iKyaGYDfw1AgB47r45z+jxm9nArLhaahSxJZrNzQnAJA7ZYqwIidqophhAx1Exdqk/byPA1vzT
+          YjpvzQa2pmxRhGIJsa33mGBvBT1wBIpNq3Di5CGhtEFla4o+PKAIxz7+9wjcJrCqWDcqoAwdJdgr
+          4yFxfccmBZwQeC+YjujhaNM5y3OrjmuUIQGtMCs3HAUFmsRH8uin8I4TjH/DEHhcdE69nsdEmbCT
+          y2YbmsruUcnZAEcbWhTgA5AKBFTFJJfC0ANFsCBURjqQs1HDGQKfIOHJJp+jJg+z2RQ+Uk3BpnAu
+          4Na3W77jt/Y8hjNh8OgzXaCyykz7Tu2ebDGbFRAofs0EQwnK3cNxdNho9qy4TdJn5+15Cu+taFbP
+          bt2vGf9i0zg3x7cOfWevqKxQ9F5vnvWsCDuy+lrzRFp1z8vf4sOtOxRvOlP4cMRkQ+ilKeZx78rv
+          VQQohFY0WUWwkDBYpSOCtM6hyM9Q8QmPmAqwIQzDct//iGUfc52EAih6cn3CibmGAyegOk8C1hh1
+          ujXb+HI/6QkPrdi8bbEN4WJ/ua1O4LJJvJcLfrMfKJJUu1wSx7wmotyYDn0ZAfzdrWj7autMk7hu
+          dKf8FWMmfHh67PnMcBkGdL68ospqwwCslsviB4Q7j2opyN2WG2ddhX4IHU6CbT3xHTC6K/vf6fyI
+          uy+dYvl/6AfA5TlFv2sSenKvSx7cEubL+V9utzZ3CRs5i2K9O1AsMTWJ+rt1aHbr+dot1svZ48qM
+          XkbfAQAA//8DABEhtZvFBQAA
       headers:
         CF-Cache-Status:
           - DYNAMIC
         CF-RAY:
-          - 8df952625d0e6802-SJC
+          - 8effb6742d1526ae-SJC
         Connection:
           - keep-alive
         Content-Encoding:
@@ -387,14 +388,14 @@ interactions:
         Content-Type:
           - application/json
         Date:
-          - Fri, 08 Nov 2024 23:25:24 GMT
+          - Tue, 10 Dec 2024 19:41:30 GMT
         Server:
           - cloudflare
         Set-Cookie:
-          - __cf_bm=h5U8y_cnXjJFRsO9gNGsRDfVg_WNTd.dD8V2ECnaukI-1731108324-1.0.1.1-TZg1ChocRLk67ht1baVmu58G2FMju83G108rLn81ugzgzE3ZsnFObSzCamjfv_DQoXJUQ49Pr4SdHHzZS3CW0A;
-            path=/; expires=Fri, 08-Nov-24 23:55:24 GMT; domain=.api.openai.com; HttpOnly;
+          - __cf_bm=Ychzvrqm8ii5aDVzJRWeETd1lUPwQFgXTorbDzGbNxQ-1733859690-1.0.1.1-GhlacpRZRjQA5bckMgo4iHyT5aFoEvsx6Cg1zdNWJbuNZGIETllv3S8pYHROiRFxmNz8.TU6FMaj9oOlXsdElg;
+            path=/; expires=Tue, 10-Dec-24 20:11:30 GMT; domain=.api.openai.com; HttpOnly;
             Secure; SameSite=None
-          - _cfuvid=seDaU9NsvV__5EgEMm84fbcz.hGsr9A6pkA0d2BYwng-1731108324458-0.0.1.1-604800000;
+          - _cfuvid=Gs6FAfK3UMx7rqMV9YrWcviJA6JZhiKo7WktWRVE4qw-1733859690932-0.0.1.1-604800000;
             path=/; domain=.api.openai.com; HttpOnly; Secure; SameSite=None
         Transfer-Encoding:
           - chunked
@@ -407,7 +408,7 @@ interactions:
         openai-organization:
           - future-house-xr4tdh
         openai-processing-ms:
-          - "2701"
+          - "1206"
         openai-version:
           - "2020-10-01"
         strict-transport-security:
@@ -419,13 +420,13 @@ interactions:
         x-ratelimit-remaining-requests:
           - "29999"
         x-ratelimit-remaining-tokens:
-          - "149999625"
+          - "149999624"
         x-ratelimit-reset-requests:
           - 2ms
         x-ratelimit-reset-tokens:
           - 0s
         x-request-id:
-          - req_2335626956aed6bc9770d9d796e88db9
+          - req_954e73c5b4c64b2d1a3b3be6fd3176ab
       status:
         code: 200
         message: OK
diff --git a/tests/cassettes/TestReActAgent.test_agent_grad[False-gpt-4o].yaml b/tests/cassettes/TestReActAgent.test_agent_grad[False-gpt-4o].yaml
index 19bfa9fa..1a0b4044 100644
--- a/tests/cassettes/TestReActAgent.test_agent_grad[False-gpt-4o].yaml
+++ b/tests/cassettes/TestReActAgent.test_agent_grad[False-gpt-4o].yaml
@@ -1,24 +1,18 @@
 interactions:
   - request:
       body:
-        '{"messages": [{"role": "system", "content": "Answer the following questions
+        '{"messages":[{"role":"system","content":"Answer the following questions
         as best you can, using the provided tools.\n\nUse the following format:\n\nThought:
         you should always think about what to do\nAction: the action to take, should
         be one of the provided tools with necessary arguments\nObservation: the result
         of the action\n... (this Thought/Action/Observation can repeat N times)\n\nExample:\n\nThought:
         I need to use the get_weather tool\nAction: get_weather(\"New York\", 7)\nObservation:
-        The 7 day forecast for New York is [...]"}, {"role": "user", "content": "Write
-        a 5 word story via print_story"}], "model": "gpt-4o", "stop": ["Observation:",
-        "Action:"], "temperature": 0.1, "tool_choice": "none", "tools": [{"type": "function",
-        "function": {"name": "print_story", "description": "Print a story.", "parameters":
-        {"type": "object", "properties": {"story": {"description": "Story to print.",
-        "title": "Story", "type": "string"}}, "required": ["story"]}}}, {"type": "function",
-        "function": {"name": "cast_float", "description": "Cast the input argument x
-        to a float.", "parameters": {"type": "object", "properties": {"x": {"title":
-        "X", "type": "string"}}, "required": ["x"]}}}, {"type": "function", "function":
-        {"name": "cast_int", "description": "Cast the input argument x to an integer.",
-        "parameters": {"type": "object", "properties": {"x": {"title": "X", "type":
-        "number"}}, "required": ["x"]}}}]}'
+        The 7 day forecast for New York is [...]"},{"role":"user","content":"Write a
+        5 word story via print_story"}],"model":"gpt-4o","n":1,"stop":["Observation:","Action:"],"temperature":0.1,"tool_choice":"none","tools":[{"type":"function","function":{"name":"print_story","description":"Print
+        a story.","parameters":{"type":"object","properties":{"story":{"description":"Story
+        to print.","title":"Story","type":"string"}},"required":["story"]}}},{"type":"function","function":{"name":"cast_float","description":"Cast
+        the input argument x to a float.","parameters":{"type":"object","properties":{"x":{"title":"X","type":"string"}},"required":["x"]}}},{"type":"function","function":{"name":"cast_int","description":"Cast
+        the input argument x to an integer.","parameters":{"type":"object","properties":{"x":{"title":"X","type":"number"}},"required":["x"]}}}]}'
       headers:
         accept:
           - application/json
@@ -27,13 +21,13 @@ interactions:
         connection:
           - keep-alive
         content-length:
-          - "1413"
+          - "1345"
         content-type:
           - application/json
         host:
           - api.openai.com
         user-agent:
-          - AsyncOpenAI/Python 1.54.3
+          - AsyncOpenAI/Python 1.56.2
         x-stainless-arch:
           - arm64
         x-stainless-async:
@@ -43,11 +37,11 @@ interactions:
         x-stainless-os:
           - MacOS
         x-stainless-package-version:
-          - 1.54.3
+          - 1.56.2
         x-stainless-raw-response:
           - "true"
         x-stainless-retry-count:
-          - "1"
+          - "0"
         x-stainless-runtime:
           - CPython
         x-stainless-runtime-version:
@@ -57,19 +51,18 @@ interactions:
     response:
       body:
         string: !!binary |
-          H4sIAAAAAAAAA4xSTYvbMBC9+1cMOseLHWezaW6lpZBDobSlhX5gtNLYVitrVGlMGpb89yLbG2fp
-          FnrRYd4H783oIQMQRos9CNVJVr23+cv3H3RR/Bre+rY4vBteGTqeXh/e9J+/VJ9YrJKC7n+g4kfV
-          jaLeW2RDboJVQMmYXMu7qiyL3Xq7HYGeNNokaz3nG8rXxXqTF7u82M7CjozCKPbwNQMAeBjfFNFp
-          /C32UKweJz3GKFsU+wsJQASyaSJkjCaydFPcGVTkGN2Y+mNHQ9vxHg5wNNbCEBG4Q/DBOK4jUzhB
-          MziVKgETTIVAwi0cKWgYGTff3LV9wGaIMrVzg7Xz/HzJa6n1ge7jjF/mjXEmdnVAGcmlbJHJixE9
-          ZwDfx70MT6oKH6j3XDP9RJcM1+V28hPLJRa0vJtBJpb2SlVVq2f8ao0sjY1XmxVKqg71Il3OIAdt
-          6ArIrlr/neY576m5ce3/2C+AUugZde0DaqOeNl5oAdNH/RftsuUxsIinyNjXjXEthvErpHs0vi5v
-          X+hdtSmVEtk5+wMAAP//AwBwQI8jNAMAAA==
+          H4sIAAAAAAAAAwAAAP//jFJNSwMxFLzvrwjv3JXtt92boqigBykVqciSJq+70WwSkqxUS/+7ZLtt
+          t7SClxxm3kxmXrKOCAHBISXACupZaWR8xZPpzfzpa3b9+vzC5qu7h/vpz+2wkr2ZeYROUOjFBzK/
+          U10wXRqJXmi1pZlF6jG4dsf9/uVwMkm6NVFqjjLIcuPjgY57SW8QJ5dxMmqEhRYMHaTkLSKEkHV9
+          hoiK4wpSknR2SInO0Rwh3Q8RAlbLgAB1TjhPlYfOgWRaeVR16jZscVk5GlKpSsoG3+zvkTo3Vi9c
+          w+/xpVDCFZlF6rQKns5rAzW7iQh5r/tURxHBWF0an3n9iSoY9rqjrR8cNnhgm67gtafynOjILuPo
+          qZCutRBglBXITxwJAVpxoVtE1Cp9Guac97a4UPl/7A8EY2g88sxY5IKdLVybh//119h+yXVgcN/O
+          Y5kthcrRGiu2T7w0GRszmuCCMgrRJvoFAAD//wMAyt7DnOsCAAA=
       headers:
         CF-Cache-Status:
           - DYNAMIC
         CF-RAY:
-          - 8df951066fb0cf22-SJC
+          - 8effbba17cc7173e-SJC
         Connection:
           - keep-alive
         Content-Encoding:
@@ -77,7 +70,7 @@ interactions:
         Content-Type:
           - application/json
         Date:
-          - Fri, 08 Nov 2024 23:24:26 GMT
+          - Tue, 10 Dec 2024 19:45:02 GMT
         Server:
           - cloudflare
         Transfer-Encoding:
@@ -91,7 +84,7 @@ interactions:
         openai-organization:
           - future-house-xr4tdh
         openai-processing-ms:
-          - "427"
+          - "401"
         openai-version:
           - "2020-10-01"
         strict-transport-security:
@@ -109,33 +102,25 @@ interactions:
         x-ratelimit-reset-tokens:
           - 0s
         x-request-id:
-          - req_5e6850c1a253c1c7006e3680cc9df401
+          - req_d795cef300521e5425ec74e7a7a21d33
       status:
         code: 200
         message: OK
   - request:
       body:
-        '{"messages": [{"role": "system", "content": "Answer the following questions
+        '{"messages":[{"role":"system","content":"Answer the following questions
         as best you can, using the provided tools.\n\nUse the following format:\n\nThought:
         you should always think about what to do\nAction: the action to take, should
         be one of the provided tools with necessary arguments\nObservation: the result
         of the action\n... (this Thought/Action/Observation can repeat N times)\n\nExample:\n\nThought:
         I need to use the get_weather tool\nAction: get_weather(\"New York\", 7)\nObservation:
-        The 7 day forecast for New York is [...]"}, {"role": "user", "content": "Write
-        a 5 word story via print_story"}, {"role": "assistant", "content": "Thought:
-        I will use the print_story function to create a 5 word story.\n. Based on this
-        reasoning, let''s select the appropriate tool!\nAction: "}, {"role": "user",
-        "content": "Continue..."}], "model": "gpt-4o", "stop": ["Observation:", "Action:"],
-        "temperature": 0.1, "tool_choice": "required", "tools": [{"type": "function",
-        "function": {"name": "print_story", "description": "Print a story.", "parameters":
-        {"type": "object", "properties": {"story": {"description": "Story to print.",
-        "title": "Story", "type": "string"}}, "required": ["story"]}}}, {"type": "function",
-        "function": {"name": "cast_float", "description": "Cast the input argument x
-        to a float.", "parameters": {"type": "object", "properties": {"x": {"title":
-        "X", "type": "string"}}, "required": ["x"]}}}, {"type": "function", "function":
-        {"name": "cast_int", "description": "Cast the input argument x to an integer.",
-        "parameters": {"type": "object", "properties": {"x": {"title": "X", "type":
-        "number"}}, "required": ["x"]}}}]}'
+        The 7 day forecast for New York is [...]"},{"role":"user","content":"Write a
+        5 word story via print_story"},{"role":"assistant","content":"Thought: . Based
+        on this reasoning, let''s select the appropriate tool!\nAction: "},{"role":"user","content":"Continue..."}],"model":"gpt-4o","n":1,"stop":["Observation:","Action:"],"temperature":0.1,"tool_choice":"required","tools":[{"type":"function","function":{"name":"print_story","description":"Print
+        a story.","parameters":{"type":"object","properties":{"story":{"description":"Story
+        to print.","title":"Story","type":"string"}},"required":["story"]}}},{"type":"function","function":{"name":"cast_float","description":"Cast
+        the input argument x to a float.","parameters":{"type":"object","properties":{"x":{"title":"X","type":"string"}},"required":["x"]}}},{"type":"function","function":{"name":"cast_int","description":"Cast
+        the input argument x to an integer.","parameters":{"type":"object","properties":{"x":{"title":"X","type":"number"}},"required":["x"]}}}]}'
       headers:
         accept:
           - application/json
@@ -144,13 +129,13 @@ interactions:
         connection:
           - keep-alive
         content-length:
-          - "1642"
+          - "1503"
         content-type:
           - application/json
         host:
           - api.openai.com
         user-agent:
-          - AsyncOpenAI/Python 1.54.3
+          - AsyncOpenAI/Python 1.56.2
         x-stainless-arch:
           - arm64
         x-stainless-async:
@@ -160,7 +145,7 @@ interactions:
         x-stainless-os:
           - MacOS
         x-stainless-package-version:
-          - 1.54.3
+          - 1.56.2
         x-stainless-raw-response:
           - "true"
         x-stainless-retry-count:
@@ -174,20 +159,20 @@ interactions:
     response:
       body:
         string: !!binary |
-          H4sIAAAAAAAAA4xTTW/bMAy9+1cIPCeF89F8+DZsK9B1ly6nrRkMVaZtpbIkSDS2JMh/HywntpN1
-          wHwQBD6+R/KJPkaMgcwgYSBKTqKyavzh2yaLZf72kG/o02PMn9ffC7/Zfy0f1vMDjBqGed2hoAvr
-          TpjKKiRpdAsLh5ywUZ0sZ5NJvJouFgGoTIaqoRWWxnMznsbT+ThejePFmVgaKdBDwl4ixhg7hrNp
-          UWf4GxIWjy6RCr3nBULSJTEGzqgmAtx76YlrglEPCqMJddO1rpUaAGSMSgVXqi/cfsfBvfeJK5Xy
-          L0+r5Wo3P8i6evoo8tnz58Pjj2o3qNdK721oKK+16PwZ4F08uSnGGGheBa51UlPqybj9DZ0x4K6o
-          K9TUtA7HLYS0LSRb2NTaI7FfpfQWnWcehUPyjAzTsijpbgsnuJI7Re/dfw6ccpjXnquzhef4qXsT
-          ZQrrzKu/sRhyqaUvU4fch1GHjkeXaqEO1FePCtaZylJK5g11Izu9X7aq0O/cAD2vB5Ahrgbx5YV1
-          pZdmSFyGV+8WTXBRYtZT+4XjdSbNAIgGs//dzXva7fxSF/8j3wNCoCXMUuswk+J64j7NYfNL/iut
-          czk0DH7vCas0l7pAF/YrLKlNJ/frbDWbT4SA6BT9AQAA//8DAPrQwbQeBAAA
+          H4sIAAAAAAAAA4xTTW/bMAy9+1cIPCeF52RN4ltXdMCwDSvW7rAthaHIjK1OX5BoNEGQ/z5YSWwn
+          7YD5IAh8fI/kE71LGANZQs5A1JyEdmp8U6YPdwv7XT3/nE1u7eabuX98/Dz9+uFF3xoYtQy7ekZB
+          J9aVsNopJGmPsPDICVvVd7PJZP5+sUizCGhbompplaPx1I6zNJuO0/k4vT4SaysFBsjZ74Qxxnbx
+          bFs0JW4gZ+noFNEYAq8Q8i6JMfBWtRHgIchA3BCMelBYQ2jark2j1AAga1UhuFJ94cO3G9x7n7hS
+          xY+5+LUw2aePZnt3f/NFbypabydGD+odpLcuNrRujOj8GeBdPL8oxhgYriPXeWmoCGT99oLOGHBf
+          NRoNta3DbgkxbQn5Eh4aE5DYSy2DQx9YQOGRAiPLjKxqulrCHs7k9slb96eBUx7XTeDqaOExvu/e
+          RNnKebsKFxbDWhoZ6sIjD3HUoePJqVqsA83Zo4LzVjsqyP5B08pm0+ygCv3ODdDjegBZ4moQvz6x
+          zvSKEonL+Ordogkuaix7ar9wvCmlHQDJYPbX3bylfZhfmup/5HtACHSEZeE8llKcT9yneWx/yX+l
+          dS7HhiFsA6Eu1tJU6ON+xSV1hZgJnuKKCw7JPvkLAAD//wMAGNjVLB4EAAA=
       headers:
         CF-Cache-Status:
           - DYNAMIC
         CF-RAY:
-          - 8df9510b297fcf22-SJC
+          - 8effbba4a856173e-SJC
         Connection:
           - keep-alive
         Content-Encoding:
@@ -195,7 +180,7 @@ interactions:
         Content-Type:
           - application/json
         Date:
-          - Fri, 08 Nov 2024 23:24:27 GMT
+          - Tue, 10 Dec 2024 19:45:02 GMT
         Server:
           - cloudflare
         Transfer-Encoding:
@@ -209,7 +194,7 @@ interactions:
         openai-organization:
           - future-house-xr4tdh
         openai-processing-ms:
-          - "432"
+          - "702"
         openai-version:
           - "2020-10-01"
         strict-transport-security:
@@ -221,13 +206,13 @@ interactions:
         x-ratelimit-remaining-requests:
           - "9999"
         x-ratelimit-remaining-tokens:
-          - "29999812"
+          - "29999828"
         x-ratelimit-reset-requests:
           - 6ms
         x-ratelimit-reset-tokens:
           - 0s
         x-request-id:
-          - req_13ac2ef5dca0872702df350e4094e09d
+          - req_c220edc212005e09ccfdb6d91be9822d
       status:
         code: 200
         message: OK
diff --git a/tests/cassettes/TestReActAgent.test_agent_grad[True-gpt-4-turbo].yaml b/tests/cassettes/TestReActAgent.test_agent_grad[True-gpt-4-turbo].yaml
index f77a5374..387c3804 100644
--- a/tests/cassettes/TestReActAgent.test_agent_grad[True-gpt-4-turbo].yaml
+++ b/tests/cassettes/TestReActAgent.test_agent_grad[True-gpt-4-turbo].yaml
@@ -1,7 +1,7 @@
 interactions:
   - request:
       body:
-        '{"messages": [{"role": "system", "content": "Answer the following questions
+        '{"messages":[{"role":"system","content":"Answer the following questions
         as best you can. You have access to the following tools:\n\nNAME: print_story\n\nSYNOPSIS:\n    print_story(string
         story)\n\nDESCRIPTION:\n    Print a story.\n\nPARAMETERS:\n    story (string):
         Story to print.\n\nNAME: cast_float\n\nSYNOPSIS:\n    cast_float(string x)\n\nDESCRIPTION:\n    Cast
@@ -14,9 +14,8 @@ interactions:
         tuple\nObservation: the result of the action\n... (this Thought/Action/Action
         Input/Observation can repeat N times)\n\nExample:\n\nThought: I need to use
         the get_weather tool\nAction: get_weather\nAction Input: \"New York\", 7\nObservation:
-        The 7 day forecast for New York is [...]"}, {"role": "user", "content": "Write
-        a 5 word story via print_story"}], "model": "gpt-4-turbo", "stop": ["Observation:"],
-        "temperature": 0.1}'
+        The 7 day forecast for New York is [...]"},{"role":"user","content":"Write a
+        5 word story via print_story"}],"model":"gpt-4-turbo","n":1,"stop":["Observation:"],"temperature":0.1}'
       headers:
         accept:
           - application/json
@@ -25,13 +24,13 @@ interactions:
         connection:
           - keep-alive
         content-length:
-          - "1289"
+          - "1281"
         content-type:
           - application/json
         host:
           - api.openai.com
         user-agent:
-          - AsyncOpenAI/Python 1.54.3
+          - AsyncOpenAI/Python 1.56.2
         x-stainless-arch:
           - arm64
         x-stainless-async:
@@ -41,7 +40,7 @@ interactions:
         x-stainless-os:
           - MacOS
         x-stainless-package-version:
-          - 1.54.3
+          - 1.56.2
         x-stainless-raw-response:
           - "true"
         x-stainless-retry-count:
@@ -55,20 +54,20 @@ interactions:
     response:
       body:
         string: !!binary |
-          H4sIAAAAAAAAA4xSy27bMBC8+ysWPLWAZci26oduORVpD0WanloXAk2uJCYUlyBXbYwg/15QfshB
-          U6CXBbizM5jZ5fMEQBgtShCqlaw6b7Obr/fq0B70g84Ls97cfdp+ueu7etl9//j4WUwTg/YPqPjM
-          minqvEU25I6wCigZk+p8vZzP881iVQxARxptojWesyLjPuwpW+SLIsuLLN+e2C0ZhVGU8GMCAPA8
-          1OTTaXwSJeTTc6fDGGWDorwMAYhANnWEjNFElo7FdAQVOUY3WP/WUt+0XMItOEQNTHA0DhJiS4Eh
-          MoUDKHJJyLgGqAZ8kortAWrzC+E3BR1nO3ejUvgSfDCOq4F2bsKt8z2X8G4n7o1Fx+BM0/IUYq81
-          OtiH9ASb6mwn3u/ctd2AdR9lWpnrrT31Xy75LTU+0D6e8Eu/Ns7EtgooI7mUNTJ5MaAvE4Cfw577
-          V6sTPlDnuWJ6RJcEF6v1UU+M5x3R5eIEMrG0V6ztdvqGXqWRpbHx6lJCSdWiHqnjWWWvDV0Bk6vU
-          f7t5S/uY3Ljmf+RHQCn0jLryAbVRrxOPYwHT7//X2GXLg2ERD5Gxq2rjGgzD/0j3qH2lP+hNsdps
-          VygmL5M/AAAA//8DAB72J5aJAwAA
+          H4sIAAAAAAAAAwAAAP//jFJda9tAEHzXr1juqQXLKLbrxHoLLaWGkkCbPNVFnO9W0tXS7XG3SmKC
+          /3s5+UM2bSAv97CzM8zM3msCIIwWOQhVS1ata9Jbnf1Ap+/n4fOXr8u7u+enh8ern2GD9033TYwi
+          g9Z/UPGRNVbUugbZkN3DyqNkjKpX19PpzafFzXzeAy1pbCKtcpzOUu78mtJJNpml2SzNFgd2TUZh
+          EDn8SgAAXvs3+rQaX0QO2eg4aTEEWaHIT0sAwlMTJ0KGYAJLy2I0gIoso+2tP9TUVTXnsASLqIEJ
+          9sZBQqjJMwQmvwVFNgoZWwGVgC9ScbOF0jwhPJPXYbyytyqGz8F5Y7noacchLK3rOIcPK/GdAsMG
+          twFK6qyGzmr0EKiU45X4uLLnPj2WXZCxK9s1zWG+OwVvqHKe1uGAn+alsSbUhUcZyMaQgcmJHt0l
+          AL/7gruLzoTz1DoumDZoo+Bkfr3XE8NdB3R6aF8wsWzOWIsj60Kv0MjSNOHsREJJVaMeqMM9ZacN
+          nQHJWep/3fxPe5/c2Oo98gOgFDpGXTiP2qjLxMOax/jt31o7tdwbFmEbGNuiNLZC33+MeI/SFVk5
+          zya4Xmspkl3yFwAA//8DACRd5fOCAwAA
       headers:
         CF-Cache-Status:
           - DYNAMIC
         CF-RAY:
-          - 8df950f96ed8faae-SJC
+          - 8effbac5be2496e1-SJC
         Connection:
           - keep-alive
         Content-Encoding:
@@ -76,14 +75,14 @@ interactions:
         Content-Type:
           - application/json
         Date:
-          - Fri, 08 Nov 2024 23:24:25 GMT
+          - Tue, 10 Dec 2024 19:44:28 GMT
         Server:
           - cloudflare
         Set-Cookie:
-          - __cf_bm=c7R6O3UrELUEXjWCIWuHSeoyVC1vPQtK74gF209bv.I-1731108265-1.0.1.1-qxdVapBa_Vl._sXTClR4FY281YRtDM4513iEjeHJQj0zbbbpo3LwsmZ12jByo0rkXaGs4naidbkg3JyMO6Uz0w;
-            path=/; expires=Fri, 08-Nov-24 23:54:25 GMT; domain=.api.openai.com; HttpOnly;
+          - __cf_bm=w4eA4W7pA8IYO1k8Lkje9HzTMqj9Y5npKaWaeWkwIFM-1733859868-1.0.1.1-oYA98x3oha4XXhNiHiGW7FWjNP9SjrC5O3AejXBmNlzFjvWoOEQsZY0fxQO_Ug.GqL9nnHTdlD74c7WY_AC9Eg;
+            path=/; expires=Tue, 10-Dec-24 20:14:28 GMT; domain=.api.openai.com; HttpOnly;
             Secure; SameSite=None
-          - _cfuvid=dNbf6A6B09bSG6qPpcCLepBZLGf76gH8k8N9P034eD0-1731108265380-0.0.1.1-604800000;
+          - _cfuvid=XW2FDZWQR_WIY.n9cSueh3jWdAwWZ17PpS5rlwQFrEc-1733859868656-0.0.1.1-604800000;
             path=/; domain=.api.openai.com; HttpOnly; Secure; SameSite=None
         Transfer-Encoding:
           - chunked
@@ -96,25 +95,25 @@ interactions:
         openai-organization:
           - future-house-xr4tdh
         openai-processing-ms:
-          - "1367"
+          - "2032"
         openai-version:
           - "2020-10-01"
         strict-transport-security:
           - max-age=31536000; includeSubDomains; preload
         x-ratelimit-limit-requests:
-          - "3000"
+          - "10000"
         x-ratelimit-limit-tokens:
-          - "250000"
+          - "2000000"
         x-ratelimit-remaining-requests:
-          - "2999"
+          - "9999"
         x-ratelimit-remaining-tokens:
-          - "249709"
+          - "1999710"
         x-ratelimit-reset-requests:
-          - 20ms
+          - 6ms
         x-ratelimit-reset-tokens:
-          - 69ms
+          - 8ms
         x-request-id:
-          - req_e76a93ae982fe7053672aa6cc9582e62
+          - req_f1e1966b85d8c1b6d48c54a361a48331
       status:
         code: 200
         message: OK
diff --git a/tests/cassettes/TestReActAgent.test_multi_step[False].yaml b/tests/cassettes/TestReActAgent.test_multi_step[False].yaml
index 11121c35..22141319 100644
--- a/tests/cassettes/TestReActAgent.test_multi_step[False].yaml
+++ b/tests/cassettes/TestReActAgent.test_multi_step[False].yaml
@@ -1,25 +1,19 @@
 interactions:
   - request:
       body:
-        '{"messages": [{"role": "system", "content": "Answer the following questions
+        '{"messages":[{"role":"system","content":"Answer the following questions
         as best you can, using the provided tools.\n\nUse the following format:\n\nThought:
         you should always think about what to do\nAction: the action to take, should
         be one of the provided tools with necessary arguments\nObservation: the result
         of the action\n... (this Thought/Action/Observation can repeat N times)\n\nExample:\n\nThought:
         I need to use the get_weather tool\nAction: get_weather(\"New York\", 7)\nObservation:
-        The 7 day forecast for New York is [...]"}, {"role": "user", "content": "Cast
-        ''5.5'' to a float, then to an integer, and finally use it to write a story
-        of that many words."}], "model": "gpt-4o-2024-08-06", "parallel_tool_calls":
-        false, "stop": ["Observation:", "Action:"], "tool_choice": "none", "tools":
-        [{"type": "function", "function": {"name": "print_story", "description": "Print
-        a story.", "parameters": {"type": "object", "properties": {"story": {"description":
-        "Story to print.", "title": "Story", "type": "string"}}, "required": ["story"]}}},
-        {"type": "function", "function": {"name": "cast_float", "description": "Cast
-        the input argument x to a float.", "parameters": {"type": "object", "properties":
-        {"x": {"title": "X", "type": "string"}}, "required": ["x"]}}}, {"type": "function",
-        "function": {"name": "cast_int", "description": "Cast the input argument x to
-        an integer.", "parameters": {"type": "object", "properties": {"x": {"title":
-        "X", "type": "number"}}, "required": ["x"]}}}]}'
+        The 7 day forecast for New York is [...]"},{"role":"user","content":"Cast ''5.5''
+        to a float, then to an integer, and finally use it to write a story of that
+        many words."}],"model":"gpt-4o-2024-08-06","n":1,"parallel_tool_calls":false,"stop":["Observation:","Action:"],"tool_choice":"none","tools":[{"type":"function","function":{"name":"print_story","description":"Print
+        a story.","parameters":{"type":"object","properties":{"story":{"description":"Story
+        to print.","title":"Story","type":"string"}},"required":["story"]}}},{"type":"function","function":{"name":"cast_float","description":"Cast
+        the input argument x to a float.","parameters":{"type":"object","properties":{"x":{"title":"X","type":"string"}},"required":["x"]}}},{"type":"function","function":{"name":"cast_int","description":"Cast
+        the input argument x to an integer.","parameters":{"type":"object","properties":{"x":{"title":"X","type":"number"}},"required":["x"]}}}]}'
       headers:
         accept:
           - application/json
@@ -28,13 +22,13 @@ interactions:
         connection:
           - keep-alive
         content-length:
-          - "1496"
+          - "1428"
         content-type:
           - application/json
         host:
           - api.openai.com
         user-agent:
-          - AsyncOpenAI/Python 1.54.3
+          - AsyncOpenAI/Python 1.56.2
         x-stainless-arch:
           - arm64
         x-stainless-async:
@@ -44,11 +38,11 @@ interactions:
         x-stainless-os:
           - MacOS
         x-stainless-package-version:
-          - 1.54.3
+          - 1.56.2
         x-stainless-raw-response:
           - "true"
         x-stainless-retry-count:
-          - "0"
+          - "1"
         x-stainless-runtime:
           - CPython
         x-stainless-runtime-version:
@@ -58,19 +52,19 @@ interactions:
     response:
       body:
         string: !!binary |
-          H4sIAAAAAAAAA4xSwWrcMBC9+ysGXXJZL/bGToJvgUIJ7SVtD4W2GEUa2+rKkqoZl7Zh/73Iu1k7
-          NIVcBJo37/HeYx4zAGG0aECoQbIag81vP3zc8/378M7c73/c7d8wlZWmtz9rmj7/EZvE8A/fUfET
-          a6v8GCyy8e4Iq4iSMamW15dlWdxc1/UMjF6jTbQ+cF75fFfsqry4yYurE3HwRiGJBr5kAACP85ss
-          Oo2/RAPF5mkyIpHsUTTnJQARvU0TIYkMsXQsNguovGN0s+tPg5/6gRu4A4eogT0oSQwX9ba+SD8J
-          nfWSoTORePvVrXUidhPJFMNN1p7mh7Mx6/sQ/QOd8PO8M87Q0EaU5F0yQeyDmNFDBvBtLmB6lkmE
-          6MfALfs9uiS4u6yPemKpfEHLqxPInqVdsepy84Jeq5GlsbSqUCipBtQLdelbTtr4FZCtUv/r5iXt
-          Y3Lj+tfIL4BSGBh1GyJqo54nXtYipov839q55dmwoN/EOLadcT3GEM3xKLrQVrXq6kqjRJEdsr8A
-          AAD//wMAmAtgtR0DAAA=
+          H4sIAAAAAAAAAwAAAP//jJI/b9swEMV3fYoDlyySodhWYmnLkABFhwJNprSFQJEniQ1FsuQJ+Qd/
+          94KyYyloCmThcL97D+8d+JoAMCVZBUz0nMTgdHYl8+/9n+uX+5u20Q/6+rb52n3b9vePz0+7S5ZG
+          hW1+o6A31UrYwWkkZc0BC4+cMLqeX242u6LcXZQTGKxEHWWdo2xrs3W+3mb5LssvjsLeKoGBVfAj
+          AQB4nd4Y0Uh8YhXk6dtkwBB4h6w6LQEwb3WcMB6CCsQNsXSGwhpCM6W+6+3Y9VTBjfKBUvgCBlEC
+          WRA8EFCPEMgr08FZsSrOIuDQastp9dMsPT22Y+Cxkhm1Ps73p5Dads7bJhz5ad4qo0Jfe+TBmhgo
+          kHVsovsE4Nd0jPFdP+a8HRzVZB/QRMP1pjj4sfn8Mz0vj5Ascb1QFdv0A79aInGlw+KcTHDRo5yl
+          8+35KJVdgGTR+t80H3kfmivTfcZ+BkKgI5S18yiVeN94XvMYf+f/1k5XngKz8BwIh7pVpkPvvDp8
+          kNbVpSxyIcsyb1iyT/4CAAD//wMAzORJUikDAAA=
       headers:
         CF-Cache-Status:
           - DYNAMIC
         CF-RAY:
-          - 8df95cfbef009651-SJC
+          - 8effbad9fb7067c4-SJC
         Connection:
           - keep-alive
         Content-Encoding:
@@ -78,15 +72,9 @@ interactions:
         Content-Type:
           - application/json
         Date:
-          - Fri, 08 Nov 2024 23:32:36 GMT
+          - Tue, 10 Dec 2024 19:44:30 GMT
         Server:
           - cloudflare
-        Set-Cookie:
-          - __cf_bm=WKotkLMRrc2Xef4Wt20THwR7B3nUMnqvEep67nlhetk-1731108756-1.0.1.1-6i3zk77orf_J7hJlSoW3c2v1yn1wRxvKWgWnrcbwUhcsEDUx5FnpaQ6ayIN6AoYYodGSIlrfv7DQv3sm_B75mA;
-            path=/; expires=Sat, 09-Nov-24 00:02:36 GMT; domain=.api.openai.com; HttpOnly;
-            Secure; SameSite=None
-          - _cfuvid=LgwtIHBxPREoiqN9o6k_ZNhA5wCEXzPApu7fgAc1UTQ-1731108756267-0.0.1.1-604800000;
-            path=/; domain=.api.openai.com; HttpOnly; Secure; SameSite=None
         Transfer-Encoding:
           - chunked
         X-Content-Type-Options:
@@ -98,7 +86,7 @@ interactions:
         openai-organization:
           - future-house-xr4tdh
         openai-processing-ms:
-          - "344"
+          - "551"
         openai-version:
           - "2020-10-01"
         strict-transport-security:
@@ -116,34 +104,27 @@ interactions:
         x-ratelimit-reset-tokens:
           - 0s
         x-request-id:
-          - req_daf830881bc887390fc604ba46cfa2fd
+          - req_f77c374c33390f9d18b0a5f3b86ac75f
       status:
         code: 200
         message: OK
   - request:
       body:
-        '{"messages": [{"role": "system", "content": "Answer the following questions
+        '{"messages":[{"role":"system","content":"Answer the following questions
         as best you can, using the provided tools.\n\nUse the following format:\n\nThought:
         you should always think about what to do\nAction: the action to take, should
         be one of the provided tools with necessary arguments\nObservation: the result
         of the action\n... (this Thought/Action/Observation can repeat N times)\n\nExample:\n\nThought:
         I need to use the get_weather tool\nAction: get_weather(\"New York\", 7)\nObservation:
-        The 7 day forecast for New York is [...]"}, {"role": "user", "content": "Cast
-        ''5.5'' to a float, then to an integer, and finally use it to write a story
-        of that many words."}, {"role": "assistant", "content": "Thought: I need to
-        cast ''5.5'' to a float first.\n. Based on this reasoning, let''s select the
-        appropriate tool!\nAction: "}, {"role": "user", "content": "Continue..."}],
-        "model": "gpt-4o-2024-08-06", "parallel_tool_calls": false, "stop": ["Observation:",
-        "Action:"], "tool_choice": "required", "tools": [{"type": "function", "function":
-        {"name": "print_story", "description": "Print a story.", "parameters": {"type":
-        "object", "properties": {"story": {"description": "Story to print.", "title":
-        "Story", "type": "string"}}, "required": ["story"]}}}, {"type": "function",
-        "function": {"name": "cast_float", "description": "Cast the input argument x
-        to a float.", "parameters": {"type": "object", "properties": {"x": {"title":
-        "X", "type": "string"}}, "required": ["x"]}}}, {"type": "function", "function":
-        {"name": "cast_int", "description": "Cast the input argument x to an integer.",
-        "parameters": {"type": "object", "properties": {"x": {"title": "X", "type":
-        "number"}}, "required": ["x"]}}}]}'
+        The 7 day forecast for New York is [...]"},{"role":"user","content":"Cast ''5.5''
+        to a float, then to an integer, and finally use it to write a story of that
+        many words."},{"role":"assistant","content":"Thought: First, I need to cast
+        the string ''5.5'' to a float.\n. Based on this reasoning, let''s select the
+        appropriate tool!\nAction: "},{"role":"user","content":"Continue..."}],"model":"gpt-4o-2024-08-06","n":1,"parallel_tool_calls":false,"stop":["Observation:","Action:"],"tool_choice":"required","tools":[{"type":"function","function":{"name":"print_story","description":"Print
+        a story.","parameters":{"type":"object","properties":{"story":{"description":"Story
+        to print.","title":"Story","type":"string"}},"required":["story"]}}},{"type":"function","function":{"name":"cast_float","description":"Cast
+        the input argument x to a float.","parameters":{"type":"object","properties":{"x":{"title":"X","type":"string"}},"required":["x"]}}},{"type":"function","function":{"name":"cast_int","description":"Cast
+        the input argument x to an integer.","parameters":{"type":"object","properties":{"x":{"title":"X","type":"number"}},"required":["x"]}}}]}'
       headers:
         accept:
           - application/json
@@ -152,13 +133,13 @@ interactions:
         connection:
           - keep-alive
         content-length:
-          - "1702"
+          - "1638"
         content-type:
           - application/json
         host:
           - api.openai.com
         user-agent:
-          - AsyncOpenAI/Python 1.54.3
+          - AsyncOpenAI/Python 1.56.2
         x-stainless-arch:
           - arm64
         x-stainless-async:
@@ -168,7 +149,7 @@ interactions:
         x-stainless-os:
           - MacOS
         x-stainless-package-version:
-          - 1.54.3
+          - 1.56.2
         x-stainless-raw-response:
           - "true"
         x-stainless-retry-count:
@@ -182,20 +163,20 @@ interactions:
     response:
       body:
         string: !!binary |
-          H4sIAAAAAAAAA4xTTW+bQBC98ytWczYR2GCn3HKKqqpN5TRVq7pCm2XA6+xXd5cmruX/HgE2YMeR
-          ymG1mjfvzczbYRcQAryAjABbU8+kEeHN8v6prr9spTf3dnP35/sy+ey2t+JZPPwUMGkY+nGDzB9Z
-          V0xLI9BzrTqYWaQeG9V4MYvj6HqRzltA6gJFQ6uMDxMdTqNpEkbXYTQ/ENeaM3SQkV8BIYTs2rNp
-          URX4AhmJJseIROdohZD1SYSA1aKJAHWOO0+Vh8kAMq08qqZrVQsxArzWImdUiKFw9+1G98EnKkT+
-          6e+/29lDqpYb+eOumn7dPNOPcvbtZlSvk96atqGyVqz3Z4T38eysGCGgqMSuoPN5KTT1Z2xCgNqq
-          lqh80znsVvCygmwF6VW6gj2cJO+DS/ffIxsslrWj4uDPIb7vDRe6MlY/ujP/oOSKu3Vukbp2jrGd
-          wbFaWwfqkxcDY7U0Pvf6CVUjO12knSoMCzWg8fwAeu2pGLE+xJMLenmBnvL2SfstYpStsRiowzbR
-          uuB6BASj2d92c0m7m5+r6n/kB4AxNB6L3FgsODudeEiz2Pxv76X1LrcNg9s6jzIvuarQGsvblYfS
-          5EnKyjQpkCIE++AVAAD//wMAouRTjfsDAAA=
+          H4sIAAAAAAAAAwAAAP//jFPbitswEH33V4h5ThavE+fity29Qktp2bK0zWJkaexoVzckuSSE/Pti
+          O2s7aQr1gxBz5pyZORofIkJAcMgIsC0NTFk5vePxd7F79/HL++Jhlyy1Kn6YZPb100+bLp5g0jBM
+          8YQsvLJumFFWYhBGdzBzSAM2qrfL2WyVrlfLuAWU4SgbWmXDdG6mSZzMp/FqGi9OxK0RDD1k5HdE
+          CCGH9mxa1Bx3kJFWpo0o9J5WCFmfRAg4I5sIUO+FD1QHmAwgMzqgbrrWtZQjIBgjc0alHAp332F0
+          H3yiUubP8/rNZ8OK2f5Pcn+v+UP69pv6pT6M6nXSe9s2VNaa9f6M8D6eXRQjBDRV2BX0IS+loeGC
+          TQhQV9UKdWg6h8MGdhvINpDepBs4wlnyMbp2fxzZ4LCsPZUnf07xY2+4NJV1pvAX/kEptPDb3CH1
+          7RxjO6PXam0dqM9eDKwzyoY8mGfUjWyyXHWqMCzUgN4uTmAwgcoRaz2fXNHLOQYq2iftt4hRtkU+
+          UIdtojUXZgREo9n/7uaadje/0NX/yA8AY2gD8tw65IKdTzykOWz+t3+l9S63DYPf+4AqL4Wu0Fkn
+          2pWH0uZrnsaMr9dxAdExegEAAP//AwB1X7QN+wMAAA==
       headers:
         CF-Cache-Status:
           - DYNAMIC
         CF-RAY:
-          - 8df95cfef9fe9651-SJC
+          - 8effbade2f5e67c4-SJC
         Connection:
           - keep-alive
         Content-Encoding:
@@ -203,7 +184,7 @@ interactions:
         Content-Type:
           - application/json
         Date:
-          - Fri, 08 Nov 2024 23:32:36 GMT
+          - Tue, 10 Dec 2024 19:44:31 GMT
         Server:
           - cloudflare
         Transfer-Encoding:
@@ -217,7 +198,7 @@ interactions:
         openai-organization:
           - future-house-xr4tdh
         openai-processing-ms:
-          - "359"
+          - "510"
         openai-version:
           - "2020-10-01"
         strict-transport-security:
@@ -229,43 +210,34 @@ interactions:
         x-ratelimit-remaining-requests:
           - "9999"
         x-ratelimit-remaining-tokens:
-          - "29999802"
+          - "29999800"
         x-ratelimit-reset-requests:
           - 6ms
         x-ratelimit-reset-tokens:
           - 0s
         x-request-id:
-          - req_d9470d313a7c0717e7dae8a16beed8ef
+          - req_a22592e960787b9681a567be8f799fb9
       status:
         code: 200
         message: OK
   - request:
       body:
-        '{"messages": [{"role": "system", "content": "Answer the following questions
+        '{"messages":[{"role":"system","content":"Answer the following questions
         as best you can, using the provided tools.\n\nUse the following format:\n\nThought:
         you should always think about what to do\nAction: the action to take, should
         be one of the provided tools with necessary arguments\nObservation: the result
         of the action\n... (this Thought/Action/Observation can repeat N times)\n\nExample:\n\nThought:
         I need to use the get_weather tool\nAction: get_weather(\"New York\", 7)\nObservation:
-        The 7 day forecast for New York is [...]"}, {"role": "user", "content": "Cast
-        ''5.5'' to a float, then to an integer, and finally use it to write a story
-        of that many words."}, {"role": "assistant", "content": "Thought: I need to
-        cast ''5.5'' to a float first.\n. Based on this reasoning, let''s select the
-        appropriate tool!\nAction: "}, {"role": "user", "content": "Continue..."}, {"role":
-        "assistant", "content": null, "function_call": null, "tool_calls": [{"id": "call_KvzG3U5nRjmXOg2PjwaIm3TA",
-        "type": "function", "function": {"arguments": "{\"x\": \"5.5\"}", "name": "cast_float"}}]},
-        {"role": "tool", "content": "Observation: 5.5", "name": "cast_float", "tool_call_id":
-        "call_KvzG3U5nRjmXOg2PjwaIm3TA"}], "model": "gpt-4o-2024-08-06", "parallel_tool_calls":
-        false, "stop": ["Observation:", "Action:"], "tool_choice": "none", "tools":
-        [{"type": "function", "function": {"name": "print_story", "description": "Print
-        a story.", "parameters": {"type": "object", "properties": {"story": {"description":
-        "Story to print.", "title": "Story", "type": "string"}}, "required": ["story"]}}},
-        {"type": "function", "function": {"name": "cast_float", "description": "Cast
-        the input argument x to a float.", "parameters": {"type": "object", "properties":
-        {"x": {"title": "X", "type": "string"}}, "required": ["x"]}}}, {"type": "function",
-        "function": {"name": "cast_int", "description": "Cast the input argument x to
-        an integer.", "parameters": {"type": "object", "properties": {"x": {"title":
-        "X", "type": "number"}}, "required": ["x"]}}}]}'
+        The 7 day forecast for New York is [...]"},{"role":"user","content":"Cast ''5.5''
+        to a float, then to an integer, and finally use it to write a story of that
+        many words."},{"role":"assistant","content":"Thought: First, I need to cast
+        the string ''5.5'' to a float.\n. Based on this reasoning, let''s select the
+        appropriate tool!\nAction: "},{"role":"user","content":"Continue..."},{"role":"assistant","content":null,"function_call":null,"tool_calls":[{"id":"call_k4uBLocb3yv2TTndW5DQmZmG","type":"function","function":{"arguments":"{\"x\":
+        \"5.5\"}","name":"cast_float"}}]},{"role":"tool","content":"Observation: 5.5","name":"cast_float","tool_call_id":"call_k4uBLocb3yv2TTndW5DQmZmG"}],"model":"gpt-4o-2024-08-06","n":1,"parallel_tool_calls":false,"stop":["Observation:","Action:"],"tool_choice":"none","tools":[{"type":"function","function":{"name":"print_story","description":"Print
+        a story.","parameters":{"type":"object","properties":{"story":{"description":"Story
+        to print.","title":"Story","type":"string"}},"required":["story"]}}},{"type":"function","function":{"name":"cast_float","description":"Cast
+        the input argument x to a float.","parameters":{"type":"object","properties":{"x":{"title":"X","type":"string"}},"required":["x"]}}},{"type":"function","function":{"name":"cast_int","description":"Cast
+        the input argument x to an integer.","parameters":{"type":"object","properties":{"x":{"title":"X","type":"number"}},"required":["x"]}}}]}'
       headers:
         accept:
           - application/json
@@ -274,13 +246,13 @@ interactions:
         connection:
           - keep-alive
         content-length:
-          - "2027"
+          - "1939"
         content-type:
           - application/json
         host:
           - api.openai.com
         user-agent:
-          - AsyncOpenAI/Python 1.54.3
+          - AsyncOpenAI/Python 1.56.2
         x-stainless-arch:
           - arm64
         x-stainless-async:
@@ -290,7 +262,7 @@ interactions:
         x-stainless-os:
           - MacOS
         x-stainless-package-version:
-          - 1.54.3
+          - 1.56.2
         x-stainless-raw-response:
           - "true"
         x-stainless-retry-count:
@@ -304,19 +276,19 @@ interactions:
     response:
       body:
         string: !!binary |
-          H4sIAAAAAAAAA4xSPW/bMBDd9SsOXLJIhmxLjqutYzrUQdoCDdpCYKiTyFjiEeSpaRH4vxeSFUtB
-          U6ALh3sffO/I5whAmEoUIJSWrDrXJu/vPh17c9D35D8cvhwPj++2dHd7n3l3+9WKeFDQwyMqflGt
-          FHWuRTY0wcqjZBxc19fb9TrdX+e7EeiownaQNY6TjJJNusmSdJ+ku0moySgMooBvEQDA83gOEW2F
-          v0QBafwy6TAE2aAoLiQA4akdJkKGYAJLyyKeQUWW0Y6pP2vqG80FfKQnYC0ZbkDLnwhKBoarfJVf
-          ARNIqFuSHMOTNkqDCZCv8hhuwCJWA2GkGx65FoxlbNCvvtvltR7rPsihte3bdpqfLj1aapynhzDh
-          l3ltrAm69CgD2SFzYHJiRE8RwI9xX/2rFQjnqXNcMh3RDobbdHf2E/MLLdDNBDKxbBfz7T5+w6+s
-          kKVpw2LjQkmlsZql8/PIvjK0AKJF67/TvOV9bm5s8z/2M6AUOsaqdB4ro143nmkehw/8L9ply2Ng
-          EX4Hxq6sjW3QO2/Of6h2ZZarOs8qlCiiU/QHAAD//wMAe66zYUwDAAA=
+          H4sIAAAAAAAAAwAAAP//jJJBb9swDIXv/hWELr3YgZvEbZLbtl4CBMM29DKsg6FItKVMlgSJ3lYU
+          +e+DnDR2sA7YRYf3kQ98pF4yAKYl2wATipPovCneyfLL4cPdp1btHrdCrb5aH3YPu88Ph/dzyfLU
+          4fYHFPTaNROu8wZJO3vCIiAnTK6394vFqlqv7m8H0DmJJrW1noqlK+blfFmUq6K8OzcqpwVGtoFv
+          GQDAy/CmEa3E32wDZf6qdBgjb5FtLkUALDiTFMZj1JG4JZaPUDhLaIepH5XrW0Ub+Oh+ASlOsAXF
+          fyLEXgiMsemNeQbBI8FNNatugByQQmiM4wTVrMphCxZRJjCUXdGkcgvaErYYZk92OkbApo88bcH2
+          xpz14yWXca0Pbh/P/KI32uqo6oA8OpsyRHKeDfSYAXwf9tdfrYT54DpPNbkfaJPholyf/Nh4sQmt
+          zpAccTPRl8v8Db9aInFt4uQCTHChUI6t47l4L7WbgGyS+u9p3vI+Jde2/R/7EQiBnlDWPqDU4jrx
+          WBYwfeh/lV22PAzM4nMk7OpG2xaDD/r0pxpfr2VVCrlel3uWHbM/AAAA//8DAJ7rg2RcAwAA
       headers:
         CF-Cache-Status:
           - DYNAMIC
         CF-RAY:
-          - 8df95d022cb59651-SJC
+          - 8effbae21a9367c4-SJC
         Connection:
           - keep-alive
         Content-Encoding:
@@ -324,7 +296,7 @@ interactions:
         Content-Type:
           - application/json
         Date:
-          - Fri, 08 Nov 2024 23:32:37 GMT
+          - Tue, 10 Dec 2024 19:44:32 GMT
         Server:
           - cloudflare
         Transfer-Encoding:
@@ -338,7 +310,7 @@ interactions:
         openai-organization:
           - future-house-xr4tdh
         openai-processing-ms:
-          - "579"
+          - "1085"
         openai-version:
           - "2020-10-01"
         strict-transport-security:
@@ -350,46 +322,37 @@ interactions:
         x-ratelimit-remaining-requests:
           - "9999"
         x-ratelimit-remaining-tokens:
-          - "29999796"
+          - "29999793"
         x-ratelimit-reset-requests:
           - 6ms
         x-ratelimit-reset-tokens:
           - 0s
         x-request-id:
-          - req_53a5f569ed5dd9384ae363644dbf2f4f
+          - req_a936bbebea97c7fc58839d6d0f1e735d
       status:
         code: 200
         message: OK
   - request:
       body:
-        '{"messages": [{"role": "system", "content": "Answer the following questions
+        '{"messages":[{"role":"system","content":"Answer the following questions
         as best you can, using the provided tools.\n\nUse the following format:\n\nThought:
         you should always think about what to do\nAction: the action to take, should
         be one of the provided tools with necessary arguments\nObservation: the result
         of the action\n... (this Thought/Action/Observation can repeat N times)\n\nExample:\n\nThought:
         I need to use the get_weather tool\nAction: get_weather(\"New York\", 7)\nObservation:
-        The 7 day forecast for New York is [...]"}, {"role": "user", "content": "Cast
-        ''5.5'' to a float, then to an integer, and finally use it to write a story
-        of that many words."}, {"role": "assistant", "content": "Thought: I need to
-        cast ''5.5'' to a float first.\n. Based on this reasoning, let''s select the
-        appropriate tool!\nAction: "}, {"role": "user", "content": "Continue..."}, {"role":
-        "assistant", "content": null, "function_call": null, "tool_calls": [{"id": "call_KvzG3U5nRjmXOg2PjwaIm3TA",
-        "type": "function", "function": {"arguments": "{\"x\": \"5.5\"}", "name": "cast_float"}}]},
-        {"role": "tool", "content": "Observation: 5.5", "name": "cast_float", "tool_call_id":
-        "call_KvzG3U5nRjmXOg2PjwaIm3TA"}, {"role": "assistant", "content": "Thought:
-        Now that I have cast ''5.5'' to a float, which is 5.5, I need to cast it to
-        an integer.\n. Based on this reasoning, let''s select the appropriate tool!\nAction:
-        "}, {"role": "user", "content": "Continue..."}], "model": "gpt-4o-2024-08-06",
-        "parallel_tool_calls": false, "stop": ["Observation:", "Action:"], "tool_choice":
-        "required", "tools": [{"type": "function", "function": {"name": "print_story",
-        "description": "Print a story.", "parameters": {"type": "object", "properties":
-        {"story": {"description": "Story to print.", "title": "Story", "type": "string"}},
-        "required": ["story"]}}}, {"type": "function", "function": {"name": "cast_float",
-        "description": "Cast the input argument x to a float.", "parameters": {"type":
-        "object", "properties": {"x": {"title": "X", "type": "string"}}, "required":
-        ["x"]}}}, {"type": "function", "function": {"name": "cast_int", "description":
-        "Cast the input argument x to an integer.", "parameters": {"type": "object",
-        "properties": {"x": {"title": "X", "type": "number"}}, "required": ["x"]}}}]}'
+        The 7 day forecast for New York is [...]"},{"role":"user","content":"Cast ''5.5''
+        to a float, then to an integer, and finally use it to write a story of that
+        many words."},{"role":"assistant","content":"Thought: First, I need to cast
+        the string ''5.5'' to a float.\n. Based on this reasoning, let''s select the
+        appropriate tool!\nAction: "},{"role":"user","content":"Continue..."},{"role":"assistant","content":null,"function_call":null,"tool_calls":[{"id":"call_k4uBLocb3yv2TTndW5DQmZmG","type":"function","function":{"arguments":"{\"x\":
+        \"5.5\"}","name":"cast_float"}}]},{"role":"tool","content":"Observation: 5.5","name":"cast_float","tool_call_id":"call_k4uBLocb3yv2TTndW5DQmZmG"},{"role":"assistant","content":"Thought:
+        Now that I have successfully cast ''5.5'' to the float 5.5, I need to cast the
+        float 5.5 to an integer.\n. Based on this reasoning, let''s select the appropriate
+        tool!\nAction: "},{"role":"user","content":"Continue..."}],"model":"gpt-4o-2024-08-06","n":1,"parallel_tool_calls":false,"stop":["Observation:","Action:"],"tool_choice":"required","tools":[{"type":"function","function":{"name":"print_story","description":"Print
+        a story.","parameters":{"type":"object","properties":{"story":{"description":"Story
+        to print.","title":"Story","type":"string"}},"required":["story"]}}},{"type":"function","function":{"name":"cast_float","description":"Cast
+        the input argument x to a float.","parameters":{"type":"object","properties":{"x":{"title":"X","type":"string"}},"required":["x"]}}},{"type":"function","function":{"name":"cast_int","description":"Cast
+        the input argument x to an integer.","parameters":{"type":"object","properties":{"x":{"title":"X","type":"number"}},"required":["x"]}}}]}'
       headers:
         accept:
           - application/json
@@ -398,13 +361,13 @@ interactions:
         connection:
           - keep-alive
         content-length:
-          - "2280"
+          - "2200"
         content-type:
           - application/json
         host:
           - api.openai.com
         user-agent:
-          - AsyncOpenAI/Python 1.54.3
+          - AsyncOpenAI/Python 1.56.2
         x-stainless-arch:
           - arm64
         x-stainless-async:
@@ -414,7 +377,7 @@ interactions:
         x-stainless-os:
           - MacOS
         x-stainless-package-version:
-          - 1.54.3
+          - 1.56.2
         x-stainless-raw-response:
           - "true"
         x-stainless-retry-count:
@@ -428,20 +391,20 @@ interactions:
     response:
       body:
         string: !!binary |
-          H4sIAAAAAAAAA4xTUW+bMBB+51dY9xyqhEIS8RatU7dp67T1aVon5JoD3Brbs481WZT/PgEJkC6T
-          xoNl3Xf3fXefj33AGMgcUgai4iRqq8LN1/vnXzfF7u5JF+/jkqroPv9y8/slutv+XMKsrTCPTyjo
-          VHUlTG0VkjS6h4VDTtiyLlbXi8V8vUpWHVCbHFVbVloKYxNG8ygO5+twfuQVlZECPaTse8AYY/vu
-          bFvUOW4hZfPZKVKj97xESIckxsAZ1UaAey89cU0wG0FhNKFuu9aNUhOAjFGZ4EqNwv23n9xHn7hS
-          2dsFfbyNNs27jaRvu/L2DS9eig+fP030euqd7RoqGi0Gfyb4EE9fiTEGmtfYC3rK5NksxwzuyqZG
-          TW3fsH+A7QOkyVVygLPEQ3Dp/mNigMOi8VwdnTnGD4PVypTWmUf/yjkopJa+yhxy300wNTI4qXU6
-          0Jy9FVhnaksZmWfULe31MupZYVylEV0sjyAZ4mpStVrPLvBlORKX3WMO+yO4qDAfS8c94k0uzQQI
-          JrP/3c0l7n5+qcv/oR8BIdAS5pl1mEtxPvGY5rD90/6VNrjcNQx+5wnrrJC6RGed7JYdCpvFiSiS
-          OEeOEByCPwAAAP//AwBsxhCV9QMAAA==
+          H4sIAAAAAAAAAwAAAP//jFNNi9swEL37V4g5x4s3cRLHty0tpZCWsqWnbjGKNHbU1VclmU0a8t+L
+          P2I72yzUByHmzbw38zQ+RYSA4JATYHsamLIyfuDJ4/M38cfL9Qe33R75IavM55f0++GQfoJZU2F2
+          v5CFS9UdM8pKDMLoDmYOacCG9X69WGTLTbaet4AyHGVTVtkQpyaeJ/M0TrI4WfWFeyMYesjJj4gQ
+          Qk7t2bSoOR4gJ8nsElHoPa0Q8iGJEHBGNhGg3gsfqA4wG0FmdEDddK1rKSdAMEYWjEo5CnffaXIf
+          faJSFunaqq/i99a8Xz2+kzv1or5kWfWxnuh11EfbNlTWmg3+TPAhnr8SIwQ0VdgJ+lCIq1n6DOqq
+          WqEOTd9weoLDE+TLu+UZrhLP0a37z4kBDsvaU9k708fPg9XSVNaZnX/lHJRCC78vHFLfTjA1Mrqo
+          tTpQX70VWGeUDUUwz6gb2sUq61hhXKURvV/1YDCByklVls5u8BUcAxXtYw77wyjbIx9Lxz2iNRdm
+          AkST2f/t5hZ3N7/Q1f/QjwBjaAPywjrkgl1PPKY5bP60t9IGl9uGwR99QFWUQlforBPtskNpiw1f
+          JoxvNskOonP0FwAA//8DAFcMvhr1AwAA
       headers:
         CF-Cache-Status:
           - DYNAMIC
         CF-RAY:
-          - 8df95d0698c99651-SJC
+          - 8effbae9f94c67c4-SJC
         Connection:
           - keep-alive
         Content-Encoding:
@@ -449,7 +412,7 @@ interactions:
         Content-Type:
           - application/json
         Date:
-          - Fri, 08 Nov 2024 23:32:38 GMT
+          - Tue, 10 Dec 2024 19:44:33 GMT
         Server:
           - cloudflare
         Transfer-Encoding:
@@ -463,7 +426,7 @@ interactions:
         openai-organization:
           - future-house-xr4tdh
         openai-processing-ms:
-          - "804"
+          - "888"
         openai-version:
           - "2020-10-01"
         strict-transport-security:
@@ -475,50 +438,38 @@ interactions:
         x-ratelimit-remaining-requests:
           - "9999"
         x-ratelimit-remaining-tokens:
-          - "29999750"
+          - "29999744"
         x-ratelimit-reset-requests:
           - 6ms
         x-ratelimit-reset-tokens:
           - 0s
         x-request-id:
-          - req_5ba19ba130b00d9cc8726cdb6aacfd79
+          - req_9f0fb599bd9aa3cbc99412fbeee4d0c8
       status:
         code: 200
         message: OK
   - request:
       body:
-        '{"messages": [{"role": "system", "content": "Answer the following questions
+        '{"messages":[{"role":"system","content":"Answer the following questions
         as best you can, using the provided tools.\n\nUse the following format:\n\nThought:
         you should always think about what to do\nAction: the action to take, should
         be one of the provided tools with necessary arguments\nObservation: the result
         of the action\n... (this Thought/Action/Observation can repeat N times)\n\nExample:\n\nThought:
         I need to use the get_weather tool\nAction: get_weather(\"New York\", 7)\nObservation:
-        The 7 day forecast for New York is [...]"}, {"role": "user", "content": "Cast
-        ''5.5'' to a float, then to an integer, and finally use it to write a story
-        of that many words."}, {"role": "assistant", "content": "Thought: I need to
-        cast ''5.5'' to a float first.\n. Based on this reasoning, let''s select the
-        appropriate tool!\nAction: "}, {"role": "user", "content": "Continue..."}, {"role":
-        "assistant", "content": null, "function_call": null, "tool_calls": [{"id": "call_KvzG3U5nRjmXOg2PjwaIm3TA",
-        "type": "function", "function": {"arguments": "{\"x\": \"5.5\"}", "name": "cast_float"}}]},
-        {"role": "tool", "content": "Observation: 5.5", "name": "cast_float", "tool_call_id":
-        "call_KvzG3U5nRjmXOg2PjwaIm3TA"}, {"role": "assistant", "content": "Thought:
-        Now that I have cast ''5.5'' to a float, which is 5.5, I need to cast it to
-        an integer.\n. Based on this reasoning, let''s select the appropriate tool!\nAction:
-        "}, {"role": "user", "content": "Continue..."}, {"role": "assistant", "content":
-        null, "function_call": null, "tool_calls": [{"id": "call_E1tLG2AuHAitYygGCafwfJOM",
-        "type": "function", "function": {"arguments": "{\"x\": 5.5}", "name": "cast_int"}}]},
-        {"role": "tool", "content": "Observation: 5", "name": "cast_int", "tool_call_id":
-        "call_E1tLG2AuHAitYygGCafwfJOM"}], "model": "gpt-4o-2024-08-06", "parallel_tool_calls":
-        false, "stop": ["Observation:", "Action:"], "tool_choice": "none", "tools":
-        [{"type": "function", "function": {"name": "print_story", "description": "Print
-        a story.", "parameters": {"type": "object", "properties": {"story": {"description":
-        "Story to print.", "title": "Story", "type": "string"}}, "required": ["story"]}}},
-        {"type": "function", "function": {"name": "cast_float", "description": "Cast
-        the input argument x to a float.", "parameters": {"type": "object", "properties":
-        {"x": {"title": "X", "type": "string"}}, "required": ["x"]}}}, {"type": "function",
-        "function": {"name": "cast_int", "description": "Cast the input argument x to
-        an integer.", "parameters": {"type": "object", "properties": {"x": {"title":
-        "X", "type": "number"}}, "required": ["x"]}}}]}'
+        The 7 day forecast for New York is [...]"},{"role":"user","content":"Cast ''5.5''
+        to a float, then to an integer, and finally use it to write a story of that
+        many words."},{"role":"assistant","content":"Thought: First, I need to cast
+        the string ''5.5'' to a float.\n. Based on this reasoning, let''s select the
+        appropriate tool!\nAction: "},{"role":"user","content":"Continue..."},{"role":"assistant","content":null,"function_call":null,"tool_calls":[{"id":"call_k4uBLocb3yv2TTndW5DQmZmG","type":"function","function":{"arguments":"{\"x\":
+        \"5.5\"}","name":"cast_float"}}]},{"role":"tool","content":"Observation: 5.5","name":"cast_float","tool_call_id":"call_k4uBLocb3yv2TTndW5DQmZmG"},{"role":"assistant","content":"Thought:
+        Now that I have successfully cast ''5.5'' to the float 5.5, I need to cast the
+        float 5.5 to an integer.\n. Based on this reasoning, let''s select the appropriate
+        tool!\nAction: "},{"role":"user","content":"Continue..."},{"role":"assistant","content":null,"function_call":null,"tool_calls":[{"id":"call_47pmPiqLoD6RBlbmwmN88gGu","type":"function","function":{"arguments":"{\"x\":
+        5.5}","name":"cast_int"}}]},{"role":"tool","content":"Observation: 5","name":"cast_int","tool_call_id":"call_47pmPiqLoD6RBlbmwmN88gGu"}],"model":"gpt-4o-2024-08-06","n":1,"parallel_tool_calls":false,"stop":["Observation:","Action:"],"tool_choice":"none","tools":[{"type":"function","function":{"name":"print_story","description":"Print
+        a story.","parameters":{"type":"object","properties":{"story":{"description":"Story
+        to print.","title":"Story","type":"string"}},"required":["story"]}}},{"type":"function","function":{"name":"cast_float","description":"Cast
+        the input argument x to a float.","parameters":{"type":"object","properties":{"x":{"title":"X","type":"string"}},"required":["x"]}}},{"type":"function","function":{"name":"cast_int","description":"Cast
+        the input argument x to an integer.","parameters":{"type":"object","properties":{"x":{"title":"X","type":"number"}},"required":["x"]}}}]}'
       headers:
         accept:
           - application/json
@@ -527,13 +478,13 @@ interactions:
         connection:
           - keep-alive
         content-length:
-          - "2595"
+          - "2491"
         content-type:
           - application/json
         host:
           - api.openai.com
         user-agent:
-          - AsyncOpenAI/Python 1.54.3
+          - AsyncOpenAI/Python 1.56.2
         x-stainless-arch:
           - arm64
         x-stainless-async:
@@ -543,7 +494,7 @@ interactions:
         x-stainless-os:
           - MacOS
         x-stainless-package-version:
-          - 1.54.3
+          - 1.56.2
         x-stainless-raw-response:
           - "true"
         x-stainless-retry-count:
@@ -557,19 +508,19 @@ interactions:
     response:
       body:
         string: !!binary |
-          H4sIAAAAAAAAA4xSwWrcMBS8+yseOu8Ge2MnW98ChRJKF9rk0NIUo0jPthpZT0jPcTZh/73Yu1lv
-          aAq96DDzZjTzpJcEQBgtShCqlaw6b5dX324ehufvX9PnT/XNx82PbvjyWV7dbjJ1YWuxGBV0/xsV
-          v6rOFHXeIhtye1oFlIyja3Z5nmXp+rJYT0RHGu0oazwvc1qu0lW+TNfL9OIgbMkojKKEnwkAwMt0
-          jhGdxidRQrp4RTqMUTYoyuMQgAhkR0TIGE1k6VgsZlKRY3RT6tuW+qblEjY0ALeS4Rpa+YjALYJx
-          jA0GeJS2RygWcA0OUQMTDMEwgoTIFLYwGG4Bn6Riu4UCBgo6nt250zsD1n2UY2XXW3vAd8cSlhof
-          6D4e+CNeG2diWwWUkdwYODJ5MbG7BODXtKz+TX/hA3WeK6YHdKPh+Yds7yfm55nZVX4gmVjaGc+z
-          YvGOX6WRpbHxZN1CSdWinqXz28heGzohkpPWf6d5z3vf3Ljmf+xnQin0jLryAbVRbxvPYwHH3/uv
-          seOWp8AibiNjV9XGNRh8MPsPVPsqL1Rd5BolimSX/AEAAP//AwCJYBnBSQMAAA==
+          H4sIAAAAAAAAAwAAAP//jFLBjtMwFLznK558blbZpqFNbwiJZUFaaRFwARS59mti6vhZ9st2q1X/
+          HSXNNq0AiYsPM2/GM89+SQCE0WINQjWSVett+lZnn+37x9t3n4q7XXb34Wv39PAx5KGsH3ffxKxX
+          0OYXKn5V3ShqvUU25E60CigZe9fbZZ6vinK1zAeiJY22l9We0wWl82y+SLNVmr0ZhQ0ZhVGs4XsC
+          APAynH1Ep/FZrCGbvSItxihrFOvzEIAIZHtEyBhNZOlYzCZSkWN0Q+ovDXV1w2t4oD1wIxnuoZFP
+          CNwgGMdYY4BiBvfgEDUwwT4YRpAQmcIBeitpnHE14LNUbA9QwJ6Cjjc/3OWdAbddlH1l11k74sdz
+          CUu1D7SJI3/Gt8aZ2FQBZSTXB45MXgzsMQH4OSyru+ovfKDWc8W0Q9cb5uXy5Cem55nYeT6STCzt
+          hC/m44qv/SqNLI2NF+sWSqoG9SSd3kZ22tAFkVy0/jPN37xPzY2r/8d+IpRCz6grH1Abdd14GgvY
+          /95/jZ23PAQW8RAZ22prXI3BB3P6QFtflbrIlC7LbCOSY/IbAAD//wMArvA5aEkDAAA=
       headers:
         CF-Cache-Status:
           - DYNAMIC
         CF-RAY:
-          - 8df95d0d1eab9651-SJC
+          - 8effbaf05e8b67c4-SJC
         Connection:
           - keep-alive
         Content-Encoding:
@@ -577,7 +528,7 @@ interactions:
         Content-Type:
           - application/json
         Date:
-          - Fri, 08 Nov 2024 23:32:39 GMT
+          - Tue, 10 Dec 2024 19:44:34 GMT
         Server:
           - cloudflare
         Transfer-Encoding:
@@ -591,7 +542,7 @@ interactions:
         openai-organization:
           - future-house-xr4tdh
         openai-processing-ms:
-          - "486"
+          - "742"
         openai-version:
           - "2020-10-01"
         strict-transport-security:
@@ -603,53 +554,41 @@ interactions:
         x-ratelimit-remaining-requests:
           - "9999"
         x-ratelimit-remaining-tokens:
-          - "29999745"
+          - "29999739"
         x-ratelimit-reset-requests:
           - 6ms
         x-ratelimit-reset-tokens:
           - 0s
         x-request-id:
-          - req_fad1c5ae2dfe136d6124a1bc3f567a70
+          - req_647d24cea03da8cc711d8d751099c15a
       status:
         code: 200
         message: OK
   - request:
       body:
-        '{"messages": [{"role": "system", "content": "Answer the following questions
+        '{"messages":[{"role":"system","content":"Answer the following questions
         as best you can, using the provided tools.\n\nUse the following format:\n\nThought:
         you should always think about what to do\nAction: the action to take, should
         be one of the provided tools with necessary arguments\nObservation: the result
         of the action\n... (this Thought/Action/Observation can repeat N times)\n\nExample:\n\nThought:
         I need to use the get_weather tool\nAction: get_weather(\"New York\", 7)\nObservation:
-        The 7 day forecast for New York is [...]"}, {"role": "user", "content": "Cast
-        ''5.5'' to a float, then to an integer, and finally use it to write a story
-        of that many words."}, {"role": "assistant", "content": "Thought: I need to
-        cast ''5.5'' to a float first.\n. Based on this reasoning, let''s select the
-        appropriate tool!\nAction: "}, {"role": "user", "content": "Continue..."}, {"role":
-        "assistant", "content": null, "function_call": null, "tool_calls": [{"id": "call_KvzG3U5nRjmXOg2PjwaIm3TA",
-        "type": "function", "function": {"arguments": "{\"x\": \"5.5\"}", "name": "cast_float"}}]},
-        {"role": "tool", "content": "Observation: 5.5", "name": "cast_float", "tool_call_id":
-        "call_KvzG3U5nRjmXOg2PjwaIm3TA"}, {"role": "assistant", "content": "Thought:
-        Now that I have cast ''5.5'' to a float, which is 5.5, I need to cast it to
-        an integer.\n. Based on this reasoning, let''s select the appropriate tool!\nAction:
-        "}, {"role": "user", "content": "Continue..."}, {"role": "assistant", "content":
-        null, "function_call": null, "tool_calls": [{"id": "call_E1tLG2AuHAitYygGCafwfJOM",
-        "type": "function", "function": {"arguments": "{\"x\": 5.5}", "name": "cast_int"}}]},
-        {"role": "tool", "content": "Observation: 5", "name": "cast_int", "tool_call_id":
-        "call_E1tLG2AuHAitYygGCafwfJOM"}, {"role": "assistant", "content": "Thought:
-        Now that I have the integer value 5, I need to write a story with exactly 5
+        The 7 day forecast for New York is [...]"},{"role":"user","content":"Cast ''5.5''
+        to a float, then to an integer, and finally use it to write a story of that
+        many words."},{"role":"assistant","content":"Thought: First, I need to cast
+        the string ''5.5'' to a float.\n. Based on this reasoning, let''s select the
+        appropriate tool!\nAction: "},{"role":"user","content":"Continue..."},{"role":"assistant","content":null,"function_call":null,"tool_calls":[{"id":"call_k4uBLocb3yv2TTndW5DQmZmG","type":"function","function":{"arguments":"{\"x\":
+        \"5.5\"}","name":"cast_float"}}]},{"role":"tool","content":"Observation: 5.5","name":"cast_float","tool_call_id":"call_k4uBLocb3yv2TTndW5DQmZmG"},{"role":"assistant","content":"Thought:
+        Now that I have successfully cast ''5.5'' to the float 5.5, I need to cast the
+        float 5.5 to an integer.\n. Based on this reasoning, let''s select the appropriate
+        tool!\nAction: "},{"role":"user","content":"Continue..."},{"role":"assistant","content":null,"function_call":null,"tool_calls":[{"id":"call_47pmPiqLoD6RBlbmwmN88gGu","type":"function","function":{"arguments":"{\"x\":
+        5.5}","name":"cast_int"}}]},{"role":"tool","content":"Observation: 5","name":"cast_int","tool_call_id":"call_47pmPiqLoD6RBlbmwmN88gGu"},{"role":"assistant","content":"Thought:
+        Now that I have the integer 5, I need to write a story containing exactly 5
         words.\n. Based on this reasoning, let''s select the appropriate tool!\nAction:
-        "}, {"role": "user", "content": "Continue..."}], "model": "gpt-4o-2024-08-06",
-        "parallel_tool_calls": false, "stop": ["Observation:", "Action:"], "tool_choice":
-        "required", "tools": [{"type": "function", "function": {"name": "print_story",
-        "description": "Print a story.", "parameters": {"type": "object", "properties":
-        {"story": {"description": "Story to print.", "title": "Story", "type": "string"}},
-        "required": ["story"]}}}, {"type": "function", "function": {"name": "cast_float",
-        "description": "Cast the input argument x to a float.", "parameters": {"type":
-        "object", "properties": {"x": {"title": "X", "type": "string"}}, "required":
-        ["x"]}}}, {"type": "function", "function": {"name": "cast_int", "description":
-        "Cast the input argument x to an integer.", "parameters": {"type": "object",
-        "properties": {"x": {"title": "X", "type": "number"}}, "required": ["x"]}}}]}'
+        "},{"role":"user","content":"Continue..."}],"model":"gpt-4o-2024-08-06","n":1,"parallel_tool_calls":false,"stop":["Observation:","Action:"],"tool_choice":"required","tools":[{"type":"function","function":{"name":"print_story","description":"Print
+        a story.","parameters":{"type":"object","properties":{"story":{"description":"Story
+        to print.","title":"Story","type":"string"}},"required":["story"]}}},{"type":"function","function":{"name":"cast_float","description":"Cast
+        the input argument x to a float.","parameters":{"type":"object","properties":{"x":{"title":"X","type":"string"}},"required":["x"]}}},{"type":"function","function":{"name":"cast_int","description":"Cast
+        the input argument x to an integer.","parameters":{"type":"object","properties":{"x":{"title":"X","type":"number"}},"required":["x"]}}}]}'
       headers:
         accept:
           - application/json
@@ -658,13 +597,13 @@ interactions:
         connection:
           - keep-alive
         content-length:
-          - "2845"
+          - "2733"
         content-type:
           - application/json
         host:
           - api.openai.com
         user-agent:
-          - AsyncOpenAI/Python 1.54.3
+          - AsyncOpenAI/Python 1.56.2
         x-stainless-arch:
           - arm64
         x-stainless-async:
@@ -674,7 +613,7 @@ interactions:
         x-stainless-os:
           - MacOS
         x-stainless-package-version:
-          - 1.54.3
+          - 1.56.2
         x-stainless-raw-response:
           - "true"
         x-stainless-retry-count:
@@ -688,20 +627,20 @@ interactions:
     response:
       body:
         string: !!binary |
-          H4sIAAAAAAAAA4xTTW/bMAy9+1cIPCdFvpP61m7YMAzYIRvWFstgKDJta5UlTaKHZEH++2A5sZ2s
-          BeqDIPDxPZJP9CFiDGQKMQNRcBKlVcO79dfnnfj9x/5Ffa8X24978f39Mnv8tP6sH2BQM8z2Fwo6
-          s26EKa1CkkY3sHDICWvV8XI6Ho9Wy/ltAEqToqppuaXhzAwno8lsOFoNR4sTsTBSoIeY/YgYY+wQ
-          zrpFneIOYjYanCMles9zhLhNYgycUXUEuPfSE9cEgw4URhPqumtdKdUDyBiVCK5UV7j5Dr175xNX
-          KrG7p3fpl4fVeu0mH8wdPbr7p8V0Ne3Va6T3NjSUVVq0/vTwNh5fFWMMNC8D1zqpKfFk3P6Kzhhw
-          l1claqpbh8MGQtoG4g18K5AJTsxzYkYzKpCVnG42cIQLkWP00v1nzx+HWeW5Ohl3ih/bl1Amt85s
-          /ZWxkEktfZE45D4M2Pc5OlcLdaC6eEqwzpSWEjLPqGvZ2fS2UYVu0zp0cloKIENc9VjzM+tCL0mR
-          uAxv3a6X4KLAtKN2a8arVJoeEPVm/7+bl7Sb+aXO3yLfAUKgJUwT6zCV4nLiLs1h/SO+lta6HBoG
-          v/eEZZJJnaMLWxVW0yazucjmsxQ5QnSM/gEAAP//AwDPufswFAQAAA==
+          H4sIAAAAAAAAAwAAAP//jFNNi9swEL37V4g5J4ubePPhW9nuIZTdQFuW3TTFKNLY0VZfSDJsCPnv
+          xUpiO2kK9UGIefPezDyN9wkhIDjkBNiWBqasHH7m6Tf15D5Wj4uvi4e3xcp9efu+eX5+dPVyBoOG
+          YTbvyMKZdceMshKDMPoIM4c0YKP6aToez+7ns2kWAWU4yoZW2TDMzHCUjrJhOhumkxNxawRDDzn5
+          mRBCyD6eTYua4wfkJB2cIwq9pxVC3iYRAs7IJgLUe+ED1QEGHciMDqibrnUtZQ8IxsiCUSm7wsdv
+          37t3PlEpi6WZ7p4mD6P3lx91iQv3qsbl62r20qt3lN7Z2FBZa9b608PbeH5VjBDQVEWudUKHwgfj
+          dld0QoC6qlaoQ9M67NcQ09aQr2GpGZLaGk0oCULhgGyptUKj98ShqDTyuzUc4ELxkNy6/+qZ5bCs
+          PZUnF0/xQ/ss0lTWmY2/chlKoYXfFg6pj9P2TU/O1WIdqC/eFawzyoYimN+oG9ksy46q0K1dh45G
+          JzCYQGWPNZkMbugVHAMV8eHbXWOUbZF31G7naM2F6QFJb/a/u7mlfZxf6Op/5DuAMbQBeWEdcsEu
+          J+7SHDZ/5b/SWpdjw+B3PqAqSqErdHHF4p7aYs7vU8bn83QDySH5AwAA//8DAGfXRGQhBAAA
       headers:
         CF-Cache-Status:
           - DYNAMIC
         CF-RAY:
-          - 8df95d118aa49651-SJC
+          - 8effbaf5daeb67c4-SJC
         Connection:
           - keep-alive
         Content-Encoding:
@@ -709,7 +648,7 @@ interactions:
         Content-Type:
           - application/json
         Date:
-          - Fri, 08 Nov 2024 23:32:39 GMT
+          - Tue, 10 Dec 2024 19:44:35 GMT
         Server:
           - cloudflare
         Transfer-Encoding:
@@ -723,7 +662,7 @@ interactions:
         openai-organization:
           - future-house-xr4tdh
         openai-processing-ms:
-          - "409"
+          - "690"
         openai-version:
           - "2020-10-01"
         strict-transport-security:
@@ -735,13 +674,13 @@ interactions:
         x-ratelimit-remaining-requests:
           - "9999"
         x-ratelimit-remaining-tokens:
-          - "29999699"
+          - "29999692"
         x-ratelimit-reset-requests:
           - 6ms
         x-ratelimit-reset-tokens:
           - 0s
         x-request-id:
-          - req_da74a5eb9bca14038fce7bd69d7a5a10
+          - req_3dc231dbb4775d7cb77a7e910a5c5910
       status:
         code: 200
         message: OK
diff --git a/tests/cassettes/TestReActAgent.test_multi_step[True].yaml b/tests/cassettes/TestReActAgent.test_multi_step[True].yaml
index 7eb97e33..d98bab28 100644
--- a/tests/cassettes/TestReActAgent.test_multi_step[True].yaml
+++ b/tests/cassettes/TestReActAgent.test_multi_step[True].yaml
@@ -1,7 +1,7 @@
 interactions:
   - request:
       body:
-        '{"messages": [{"role": "system", "content": "Answer the following questions
+        '{"messages":[{"role":"system","content":"Answer the following questions
         as best you can. You have access to the following tools:\n\nNAME: print_story\n\nSYNOPSIS:\n    print_story(string
         story)\n\nDESCRIPTION:\n    Print a story.\n\nPARAMETERS:\n    story (string):
         Story to print.\n\nNAME: cast_float\n\nSYNOPSIS:\n    cast_float(string x)\n\nDESCRIPTION:\n    Cast
@@ -14,9 +14,9 @@ interactions:
         tuple\nObservation: the result of the action\n... (this Thought/Action/Action
         Input/Observation can repeat N times)\n\nExample:\n\nThought: I need to use
         the get_weather tool\nAction: get_weather\nAction Input: \"New York\", 7\nObservation:
-        The 7 day forecast for New York is [...]"}, {"role": "user", "content": "Cast
-        ''5.5'' to a float, then to an integer, and finally use it to write a story
-        of that many words."}], "model": "gpt-4o-2024-08-06", "stop": ["Observation:"]}'
+        The 7 day forecast for New York is [...]"},{"role":"user","content":"Cast ''5.5''
+        to a float, then to an integer, and finally use it to write a story of that
+        many words."}],"model":"gpt-4o-2024-08-06","n":1,"stop":["Observation:"]}'
       headers:
         accept:
           - application/json
@@ -25,13 +25,13 @@ interactions:
         connection:
           - keep-alive
         content-length:
-          - "1337"
+          - "1331"
         content-type:
           - application/json
         host:
           - api.openai.com
         user-agent:
-          - AsyncOpenAI/Python 1.54.3
+          - AsyncOpenAI/Python 1.56.2
         x-stainless-arch:
           - arm64
         x-stainless-async:
@@ -41,11 +41,11 @@ interactions:
         x-stainless-os:
           - MacOS
         x-stainless-package-version:
-          - 1.54.3
+          - 1.56.2
         x-stainless-raw-response:
           - "true"
         x-stainless-retry-count:
-          - "1"
+          - "0"
         x-stainless-runtime:
           - CPython
         x-stainless-runtime-version:
@@ -55,19 +55,19 @@ interactions:
     response:
       body:
         string: !!binary |
-          H4sIAAAAAAAAA4yST2+cMBDF73yKkS9pJVgBu5ul3PZUbS9VuqsqUlMhxx7AKdiuPVSNov3uldk/
-          EDWVeuHwfn6PeWO/RABMSVYCEy0n0dsu2X7Zy+wXffz8fH/3k9zubnN/2G+/PjX74tOBxcFhHp9Q
-          0MW1EKa3HZIy+oSFQ04YUrPNMsvSIr/djKA3ErtgaywlK5Pkab5K0iJJb8/G1iiBnpXwLQIAeBm/
-          YUQt8TcrIY0vSo/e8wZZeT0EwJzpgsK498oT18TiCQqjCfU49aE1Q9NSCTvQiBLIgOCegFoET07p
-          Bm7Wi/VNABzqznCCWjlPiwe9FaFoORqqEV002Gk7UAnvRm/8/kHPf++wHjwP7fXQdWf9eO3TmcY6
-          8+jP/KrXSivfVg65NzrM7slYNtJjBPB93NvwahXMOtNbqsj8QB0C82J5ymPTTU10mZ8hGeLdTM/W
-          8Rt5lUTiqvOzzTPBRYtysk7XxAepzAxEs9Z/T/NW9qm50s3/xE9ACLSEsrIOpRKvG0/HHIaH/K9j
-          1y2PAzP/7An7qla6QWedOr2l2lbZ+oMslqtMCBYdoz8AAAD//wMAtQxDr1QDAAA=
+          H4sIAAAAAAAAA4xSTWvcMBS8+1c8dEkL9uLdjTe7vgVCIYdCUxba0C1GKz3bamVJlZ5pStj/XuT9
+          sENTyEWHmTfDzHt6TgCYkqwEJlpOonM6u5X5ZyweHraP+y9f1fxJf3z88Kn9Je62v/0NS6PC7n+g
+          oLNqJmznNJKy5kgLj5wwus5vlst1sVmvVgPRWYk6yhpH2bXNFvniOsvXWb46CVurBAZWwrcEAOB5
+          eGNEI/GJlZCnZ6TDEHiDrLwMATBvdUQYD0EF4oZYOpLCGkIzpN62tm9aKuEeDKIEslArHwgEDwTU
+          IgTyyjRwVcyKq0hzqLXlNNuZWxGLlsNoNYBnDO6N66mEdztWzIodS9/vzDSAx7oPPPY3vdYn/HBp
+          pG3jvN2HE3/Ba2VUaCuPPFgT0weyjg3sIQH4Pmyuf7EM5rztHFVkf6KJhov18ujHxluN7HJxIskS
+          1xN8XqSv+FUSiSsdJrtngosW5SgdD8V7qeyESCat/03zmvexuTLNW+xHQgh0hLJyHqUSLxuPYx7j
+          V/7f2GXLQ2AW/gTCrqqVadA7r46/qXbVRha5kJtNvmfJIfkLAAD//wMA5RGCp1YDAAA=
       headers:
         CF-Cache-Status:
           - DYNAMIC
         CF-RAY:
-          - 8df9510c9f0c22ba-SJC
+          - 8effbac5baf7d039-SJC
         Connection:
           - keep-alive
         Content-Encoding:
@@ -75,9 +75,15 @@ interactions:
         Content-Type:
           - application/json
         Date:
-          - Fri, 08 Nov 2024 23:24:27 GMT
+          - Tue, 10 Dec 2024 19:44:27 GMT
         Server:
           - cloudflare
+        Set-Cookie:
+          - __cf_bm=r6dZddUDWfN_Os.WIvmSRteRx9TQDrAqu8ra.Qbl9Rg-1733859867-1.0.1.1-bHAbl3a2MCrOO2KjtKMQsO2qa26vpz8SS4okiAhw16opeBH13XEmOd8TzokuXcMbLWibxkSbI.69x8G.cQ2EHA;
+            path=/; expires=Tue, 10-Dec-24 20:14:27 GMT; domain=.api.openai.com; HttpOnly;
+            Secure; SameSite=None
+          - _cfuvid=gioa0R.ivKqZKaFec_dzeeGl_QNH0QCLOItZoOpXstA-1733859867523-0.0.1.1-604800000;
+            path=/; domain=.api.openai.com; HttpOnly; Secure; SameSite=None
         Transfer-Encoding:
           - chunked
         X-Content-Type-Options:
@@ -89,7 +95,7 @@ interactions:
         openai-organization:
           - future-house-xr4tdh
         openai-processing-ms:
-          - "652"
+          - "895"
         openai-version:
           - "2020-10-01"
         strict-transport-security:
@@ -107,13 +113,13 @@ interactions:
         x-ratelimit-reset-tokens:
           - 0s
         x-request-id:
-          - req_fab28d5f04ad57fc92dc973d8cfb2413
+          - req_5fc910e91f3fcf4bd0193ba919285ab0
       status:
         code: 200
         message: OK
   - request:
       body:
-        '{"messages": [{"role": "system", "content": "Answer the following questions
+        '{"messages":[{"role":"system","content":"Answer the following questions
         as best you can. You have access to the following tools:\n\nNAME: print_story\n\nSYNOPSIS:\n    print_story(string
         story)\n\nDESCRIPTION:\n    Print a story.\n\nPARAMETERS:\n    story (string):
         Story to print.\n\nNAME: cast_float\n\nSYNOPSIS:\n    cast_float(string x)\n\nDESCRIPTION:\n    Cast
@@ -126,12 +132,11 @@ interactions:
         tuple\nObservation: the result of the action\n... (this Thought/Action/Action
         Input/Observation can repeat N times)\n\nExample:\n\nThought: I need to use
         the get_weather tool\nAction: get_weather\nAction Input: \"New York\", 7\nObservation:
-        The 7 day forecast for New York is [...]"}, {"role": "user", "content": "Cast
-        ''5.5'' to a float, then to an integer, and finally use it to write a story
-        of that many words."}, {"role": "assistant", "content": "Thought: I need to
-        cast the string ''5.5'' to a float first.\nAction: cast_float\nAction Input:
-        (''5.5'',)"}, {"role": "user", "content": "Observation: 5.5"}], "model": "gpt-4o-2024-08-06",
-        "stop": ["Observation:"]}'
+        The 7 day forecast for New York is [...]"},{"role":"user","content":"Cast ''5.5''
+        to a float, then to an integer, and finally use it to write a story of that
+        many words."},{"role":"assistant","content":"Thought: I need to first cast the
+        string ''5.5'' to a float.\nAction: cast_float\nAction Input: (\"5.5\",)"},{"role":"user","content":"Observation:
+        5.5"}],"model":"gpt-4o-2024-08-06","n":1,"stop":["Observation:"]}'
       headers:
         accept:
           - application/json
@@ -140,13 +145,13 @@ interactions:
         connection:
           - keep-alive
         content-length:
-          - "1526"
+          - "1514"
         content-type:
           - application/json
         host:
           - api.openai.com
         user-agent:
-          - AsyncOpenAI/Python 1.54.3
+          - AsyncOpenAI/Python 1.56.2
         x-stainless-arch:
           - arm64
         x-stainless-async:
@@ -156,7 +161,7 @@ interactions:
         x-stainless-os:
           - MacOS
         x-stainless-package-version:
-          - 1.54.3
+          - 1.56.2
         x-stainless-raw-response:
           - "true"
         x-stainless-retry-count:
@@ -170,19 +175,19 @@ interactions:
     response:
       body:
         string: !!binary |
-          H4sIAAAAAAAAA4xSTYvbMBS8+1c8dGohDnYS58O3pZCytJTS7G1bjFZ6trWVJVV6ZluW/Pci58MO
-          baEXHWbejGae9JoAMCVZCUy0nETndHr35SBzsX+Q+bp9PnSfXzYffuzfvV/x/cfdgc2iwj49o6CL
-          ai5s5zSSsuZEC4+cMLrmm2WeZ9vFejMQnZWoo6xxlK5susgWqzTbptn6LGytEhhYCY8JAMDrcMaI
-          RuJPVkI2uyAdhsAbZOV1CIB5qyPCeAgqEDfEZiMprCE0Q+qH1vZNSyV8si9wDwZRAlkQPBBQi1Br
-          ywmKeRFRbkAZwgb9/Ku5E7FmOYxWytAFgXvjeirhTTEvZm+n13qs+8Bja9NrfcaP1x7aNs7bp3Dm
-          r3itjApt5ZEHa2LmQNaxgT0mAN+GffU3K2DO285RRfY7mmi4XOxOfmx8oQl7XiYjS1xP8OKiuvGr
-          JBJXOkw2zgQXLcpROj4P76WyEyKZtP4zzd+8T82Vaf7HfiSEQEcoK+dRKnHbeBzzGD/wv8auWx4C
-          s/ArEHZVrUyD3nl1+kO1q/JiJ7fLVS4ES47JbwAAAP//AwClrP+KTAMAAA==
+          H4sIAAAAAAAAAwAAAP//jJJNa9wwEIbv/hWDTi3sLu5+r2+BUthLD20oLU0xWmlsq5U1ijRuG8L+
+          9yLvhx2aQC7GvM/My7wzeswAhNGiAKEayar1dnqj809V83B//0HHfW0Pn7/c0pY2X/ff3ru1mKQO
+          OvxExZeumaLWW2RD7oRVQMmYXN9tFovtarddb3rQkkab2mrP0yVN5/l8Oc230/zsqxoyCqMo4HsG
+          APDYf9OITuNfUUA+uSgtxihrFMW1CEAEskkRMkYTWToWkwEqcoyun/q2oa5uuICP9Ae4kQx7aORv
+          BG4QKkuSYTVbTWAPDlEDEygZGQynX+nAOMYaw+zO3agUu+h5aRxfFNg733EBb5LP2/EYAasuyrQF
+          11l71o/XXJZqH+gQz/yqV8aZ2JQBZSSXMkQmL3p6zAB+9PvrnqxE+ECt55LpF7pkuJjvTn5iuNiI
+          rs6QiaUd6evl5Bm/UiNLY+PoAkJJ1aAeWodzyU4bGoFslPr/aZ7zPiU3rn6N/QCUQs+oSx9QG/U0
+          8VAWMD3ol8quW+4HFvEhMrZlZVyNwQdzelOVL3d6lSu92+UHkR2zfwAAAP//AwBU5kw4XAMAAA==
       headers:
         CF-Cache-Status:
           - DYNAMIC
         CF-RAY:
-          - 8df951116d5022ba-SJC
+          - 8effbacc28d6d039-SJC
         Connection:
           - keep-alive
         Content-Encoding:
@@ -190,7 +195,7 @@ interactions:
         Content-Type:
           - application/json
         Date:
-          - Fri, 08 Nov 2024 23:24:28 GMT
+          - Tue, 10 Dec 2024 19:44:28 GMT
         Server:
           - cloudflare
         Transfer-Encoding:
@@ -204,7 +209,7 @@ interactions:
         openai-organization:
           - future-house-xr4tdh
         openai-processing-ms:
-          - "647"
+          - "625"
         openai-version:
           - "2020-10-01"
         strict-transport-security:
@@ -222,13 +227,13 @@ interactions:
         x-ratelimit-reset-tokens:
           - 0s
         x-request-id:
-          - req_c798eb70cb8171a0bfb79f4535ae68b9
+          - req_21906eb43abccec10d343df6ef16ef82
       status:
         code: 200
         message: OK
   - request:
       body:
-        '{"messages": [{"role": "system", "content": "Answer the following questions
+        '{"messages":[{"role":"system","content":"Answer the following questions
         as best you can. You have access to the following tools:\n\nNAME: print_story\n\nSYNOPSIS:\n    print_story(string
         story)\n\nDESCRIPTION:\n    Print a story.\n\nPARAMETERS:\n    story (string):
         Story to print.\n\nNAME: cast_float\n\nSYNOPSIS:\n    cast_float(string x)\n\nDESCRIPTION:\n    Cast
@@ -241,14 +246,13 @@ interactions:
         tuple\nObservation: the result of the action\n... (this Thought/Action/Action
         Input/Observation can repeat N times)\n\nExample:\n\nThought: I need to use
         the get_weather tool\nAction: get_weather\nAction Input: \"New York\", 7\nObservation:
-        The 7 day forecast for New York is [...]"}, {"role": "user", "content": "Cast
-        ''5.5'' to a float, then to an integer, and finally use it to write a story
-        of that many words."}, {"role": "assistant", "content": "Thought: I need to
-        cast the string ''5.5'' to a float first.\nAction: cast_float\nAction Input:
-        (''5.5'',)"}, {"role": "user", "content": "Observation: 5.5"}, {"role": "assistant",
-        "content": "Thought: Now I need to cast the float 5.5 to an integer.\nAction:
-        cast_int\nAction Input: (5.5,)"}, {"role": "user", "content": "Observation:
-        5"}], "model": "gpt-4o-2024-08-06", "stop": ["Observation:"]}'
+        The 7 day forecast for New York is [...]"},{"role":"user","content":"Cast ''5.5''
+        to a float, then to an integer, and finally use it to write a story of that
+        many words."},{"role":"assistant","content":"Thought: I need to first cast the
+        string ''5.5'' to a float.\nAction: cast_float\nAction Input: (\"5.5\",)"},{"role":"user","content":"Observation:
+        5.5"},{"role":"assistant","content":"Thought: Now that I have the float 5.5,
+        I need to cast it to an integer.\nAction: cast_int\nAction Input: (5.5,)"},{"role":"user","content":"Observation:
+        5"}],"model":"gpt-4o-2024-08-06","n":1,"stop":["Observation:"]}'
       headers:
         accept:
           - application/json
@@ -257,13 +261,13 @@ interactions:
         connection:
           - keep-alive
         content-length:
-          - "1707"
+          - "1703"
         content-type:
           - application/json
         host:
           - api.openai.com
         user-agent:
-          - AsyncOpenAI/Python 1.54.3
+          - AsyncOpenAI/Python 1.56.2
         x-stainless-arch:
           - arm64
         x-stainless-async:
@@ -273,7 +277,7 @@ interactions:
         x-stainless-os:
           - MacOS
         x-stainless-package-version:
-          - 1.54.3
+          - 1.56.2
         x-stainless-raw-response:
           - "true"
         x-stainless-retry-count:
@@ -287,20 +291,20 @@ interactions:
     response:
       body:
         string: !!binary |
-          H4sIAAAAAAAAA4xSy27bMBC86ysWPLWAbEh+R7cccynQNgUC1IVAk2uJrsRlyVUdN/C/F5QfctAU
-          yIWHmZ3BzC5fEgBhtChAqFqyal0zuv/yVU++Led6t5lUv6ZPn7s/eVerTf5k5lKkUUGbHSq+qMaK
-          WtcgG7InWnmUjNE1X07zPFtNFqueaEljE2WV49GMRpNsMhtlq1G2OAtrMgqDKOB7AgDw0r8xotX4
-          LArI0gvSYgiyQlFchwCEpyYiQoZgAkvLIh1IRZbR9qkfa+qqmgv4RPsUHsAiamCCvTeMICEw+QNw
-          LRlMAHyWipsDzGFPXgdoyFbjtb1XsXEBzhvLZS+5gPBgXccFfFiLxxpBSYZd1zrUQL/RA9cILZEd
-          r0X68Taix20XZNyQ7ZrmjB+vnRuqnKdNOPNXfGusCXXpUQaysV9gcqJnjwnAj3633at1CeepdVwy
-          /UQbDafL/OQnhmvesPMzycSyGfBZtkjf8Cs1sjRNuLmOUFLVqAfpcErZaUM3RHLT+t80b3mfmhtb
-          vcd+IJRCx6hL51Eb9brxMOYxfvb/jV233AcW4RAY23JrbIW+/xnxHltX5vM7vZrOcqVEckz+AgAA
-          //8DAKIozat4AwAA
+          H4sIAAAAAAAAAwAAAP//jFNNj9owEL3nV4x8aqWAsgsskBu3UlVtVe2hUllFxh4SU8fj2hNYuuK/
+          VwkfYdWt1EsO7817evPGeUkAhNEiB6Eqyar2drDQ2bdyPJ/W9PHX9PsHre1uQb/11+1i++lRpK2C
+          1ltUfFENFdXeIhtyJ1oFlIyt6910NJpN5rOHWUfUpNG2stLzYEyD++x+PMhmg+zhLKzIKIwihx8J
+          AMBL920jOo3PIocsvSA1xihLFPl1CEAEsi0iZIwmsnQs0p5U5Bhdl/qxoqasOIfPtAeuJMMSKrlD
+          4ArBOMYSA+ykbRAmKSxhb6yFfTCMICEyhcNJZSLgs1RsDzCBPQUdwZIrhyu3UG0dOfhgHBed5ALC
+          0vmGc3i3El+cQmg8OZDApsYUKum9cRjjcCXe36YPuGmibMtzjbVn/Hitw1LpA63jmb/iG+NMrIqA
+          MpJrV49MXnTsMQF46mpvXjUpfKDac8H0E11rOJo+nPxEf+ieHd+dSSaW9hafpm/4FRpZGhtvDieU
+          VBXqXtpfWTba0A2R3Gz9d5q3vE+bG1f+j31PKIWeURc+oDbq9cb9WMD2P/jX2LXlLrCIh8hYFxvj
+          Sgzdu2jvsfHFXE8ypefzbC2SY/IHAAD//wMAOh6TDpMDAAA=
       headers:
         CF-Cache-Status:
           - DYNAMIC
         CF-RAY:
-          - 8df951162b0c22ba-SJC
+          - 8effbad0cd60d039-SJC
         Connection:
           - keep-alive
         Content-Encoding:
@@ -308,7 +312,7 @@ interactions:
         Content-Type:
           - application/json
         Date:
-          - Fri, 08 Nov 2024 23:24:29 GMT
+          - Tue, 10 Dec 2024 19:44:29 GMT
         Server:
           - cloudflare
         Transfer-Encoding:
@@ -322,7 +326,7 @@ interactions:
         openai-organization:
           - future-house-xr4tdh
         openai-processing-ms:
-          - "688"
+          - "888"
         openai-version:
           - "2020-10-01"
         strict-transport-security:
@@ -334,13 +338,13 @@ interactions:
         x-ratelimit-remaining-requests:
           - "9999"
         x-ratelimit-remaining-tokens:
-          - "29999634"
+          - "29999630"
         x-ratelimit-reset-requests:
           - 6ms
         x-ratelimit-reset-tokens:
           - 0s
         x-request-id:
-          - req_9f63fa5e62d5dc9d364dfe622359f34a
+          - req_d1f9a64806371bef60e98109cc7df15e
       status:
         code: 200
         message: OK
diff --git a/tests/cassettes/TestReActAgent.test_react_dummyenv[False-gpt-4o].yaml b/tests/cassettes/TestReActAgent.test_react_dummyenv[False-gpt-4o].yaml
index 10a596d7..ae04e3a1 100644
--- a/tests/cassettes/TestReActAgent.test_react_dummyenv[False-gpt-4o].yaml
+++ b/tests/cassettes/TestReActAgent.test_react_dummyenv[False-gpt-4o].yaml
@@ -1,24 +1,18 @@
 interactions:
   - request:
       body:
-        '{"messages": [{"role": "system", "content": "Answer the following questions
+        '{"messages":[{"role":"system","content":"Answer the following questions
         as best you can, using the provided tools.\n\nUse the following format:\n\nThought:
         you should always think about what to do\nAction: the action to take, should
         be one of the provided tools with necessary arguments\nObservation: the result
         of the action\n... (this Thought/Action/Observation can repeat N times)\n\nExample:\n\nThought:
         I need to use the get_weather tool\nAction: get_weather(\"New York\", 7)\nObservation:
-        The 7 day forecast for New York is [...]"}, {"role": "user", "content": "Write
-        a 5 word story via print_story"}], "model": "gpt-4o", "stop": ["Observation:",
-        "Action:"], "temperature": 0.1, "tool_choice": "none", "tools": [{"type": "function",
-        "function": {"name": "print_story", "description": "Print a story.", "parameters":
-        {"type": "object", "properties": {"story": {"description": "Story to print.",
-        "title": "Story", "type": "string"}}, "required": ["story"]}}}, {"type": "function",
-        "function": {"name": "cast_float", "description": "Cast the input argument x
-        to a float.", "parameters": {"type": "object", "properties": {"x": {"title":
-        "X", "type": "string"}}, "required": ["x"]}}}, {"type": "function", "function":
-        {"name": "cast_int", "description": "Cast the input argument x to an integer.",
-        "parameters": {"type": "object", "properties": {"x": {"title": "X", "type":
-        "number"}}, "required": ["x"]}}}]}'
+        The 7 day forecast for New York is [...]"},{"role":"user","content":"Write a
+        5 word story via print_story"}],"model":"gpt-4o","n":1,"stop":["Observation:","Action:"],"temperature":0.1,"tool_choice":"none","tools":[{"type":"function","function":{"name":"print_story","description":"Print
+        a story.","parameters":{"type":"object","properties":{"story":{"description":"Story
+        to print.","title":"Story","type":"string"}},"required":["story"]}}},{"type":"function","function":{"name":"cast_float","description":"Cast
+        the input argument x to a float.","parameters":{"type":"object","properties":{"x":{"title":"X","type":"string"}},"required":["x"]}}},{"type":"function","function":{"name":"cast_int","description":"Cast
+        the input argument x to an integer.","parameters":{"type":"object","properties":{"x":{"title":"X","type":"number"}},"required":["x"]}}}]}'
       headers:
         accept:
           - application/json
@@ -27,13 +21,13 @@ interactions:
         connection:
           - keep-alive
         content-length:
-          - "1413"
+          - "1345"
         content-type:
           - application/json
         host:
           - api.openai.com
         user-agent:
-          - AsyncOpenAI/Python 1.54.3
+          - AsyncOpenAI/Python 1.56.2
         x-stainless-arch:
           - arm64
         x-stainless-async:
@@ -43,7 +37,7 @@ interactions:
         x-stainless-os:
           - MacOS
         x-stainless-package-version:
-          - 1.54.3
+          - 1.56.2
         x-stainless-raw-response:
           - "true"
         x-stainless-retry-count:
@@ -57,19 +51,18 @@ interactions:
     response:
       body:
         string: !!binary |
-          H4sIAAAAAAAAA4xSTW/UMBC951eMfN5Um+xH09x6QIgLQrQcEKDItSeJi+Ox7ImWUu1/R85uN6ko
-          Ehcf5n3ovRk/ZwDCaFGDUL1kNXib336+U7/l+850n6pt9fEr3t3fPpae7Jf9u0qskoIeHlHxi+pK
-          0eAtsiF3glVAyZhci+tNUayrcr+bgIE02iTrPOdbyst1uc3XVb7en4U9GYVR1PAtAwB4nt4U0Wn8
-          JWpYr14mA8YoOxT1hQQgAtk0ETJGE1k6FqsZVOQY3ZT6vqex67mGD3Aw1sIYEbhH8ME4biJTeIJ2
-          dCpVAiY4BMMIEnb5gYKGiXD13S3dA7ZjlKmcG609z4+XuJY6H+ghnvHLvDXOxL4JKCO5FC0yeTGh
-          xwzgx7SW8VVT4QMNnhumn+iSYVnsT35iPsSMFtdnkImlXag2m9Ubfo1GlsbGxWKFkqpHPUvnK8hR
-          G1oA2aL132ne8j41N677H/sZUAo9o258QG3U68YzLWD6p/+iXbY8BRbxKTIOTWtch2H6CekerW+K
-          3Y2uNttCKZEdsz8AAAD//wMAni45nzMDAAA=
+          H4sIAAAAAAAAAwAAAP//jFLLTsMwELznK6w9Nyh9t7kBF46o4oBAKHLtTeLieC3bQbRV/x05LX2o
+          ReLiw8zOeGbtbcIYKAk5A1HzIBqr03uZLerNyD4+NJq+NqvN4mX9Wg7fJs9y/AS9qKDlCkX4Vd0J
+          aqzGoMjsaeGQB4yu/elwOBvPZ5N5RzQkUUdZZUM6onSQDUZpNkuzyUFYkxLoIWfvCWOMbbszRjQS
+          vyFnWe8XadB7XiHkxyHGwJGOCHDvlQ/cBOidSEEmoOlSn8MOy9bzmMq0Wh/w3fEeTZV1tPQH/oiX
+          yihfFw65JxM9fSALHbtLGPvo+rQXEcE6amwoAn2iiYaD/mTvB6cNnthDVwgUuL4lurArJAautD9b
+          CAguapRXjowBb6WiMyI5K30d5pb3vrgy1X/sT4QQaAPKwjqUStws3JnH//XX2HHJXWDwax+wKUpl
+          KnTWqf0Tl7YQU8EzXHLBIdklPwAAAP//AwBFYrKZ6wIAAA==
       headers:
         CF-Cache-Status:
           - DYNAMIC
         CF-RAY:
-          - 8df951038956ce5c-SJC
+          - 8effbad2dcbc9e56-SJC
         Connection:
           - keep-alive
         Content-Encoding:
@@ -77,7 +70,7 @@ interactions:
         Content-Type:
           - application/json
         Date:
-          - Fri, 08 Nov 2024 23:24:25 GMT
+          - Tue, 10 Dec 2024 19:44:29 GMT
         Server:
           - cloudflare
         Transfer-Encoding:
@@ -91,7 +84,7 @@ interactions:
         openai-organization:
           - future-house-xr4tdh
         openai-processing-ms:
-          - "367"
+          - "896"
         openai-version:
           - "2020-10-01"
         strict-transport-security:
@@ -109,33 +102,25 @@ interactions:
         x-ratelimit-reset-tokens:
           - 0s
         x-request-id:
-          - req_a8309d869726cb11a8aad907378120f7
+          - req_6b16e63e79e61ae7b962cfe7d7a4aac2
       status:
         code: 200
         message: OK
   - request:
       body:
-        '{"messages": [{"role": "system", "content": "Answer the following questions
+        '{"messages":[{"role":"system","content":"Answer the following questions
         as best you can, using the provided tools.\n\nUse the following format:\n\nThought:
         you should always think about what to do\nAction: the action to take, should
         be one of the provided tools with necessary arguments\nObservation: the result
         of the action\n... (this Thought/Action/Observation can repeat N times)\n\nExample:\n\nThought:
         I need to use the get_weather tool\nAction: get_weather(\"New York\", 7)\nObservation:
-        The 7 day forecast for New York is [...]"}, {"role": "user", "content": "Write
-        a 5 word story via print_story"}, {"role": "assistant", "content": "Thought:
-        I will use the print_story function to write a 5-word story.\n. Based on this
-        reasoning, let''s select the appropriate tool!\nAction: "}, {"role": "user",
-        "content": "Continue..."}], "model": "gpt-4o", "stop": ["Observation:", "Action:"],
-        "temperature": 0.1, "tool_choice": "required", "tools": [{"type": "function",
-        "function": {"name": "print_story", "description": "Print a story.", "parameters":
-        {"type": "object", "properties": {"story": {"description": "Story to print.",
-        "title": "Story", "type": "string"}}, "required": ["story"]}}}, {"type": "function",
-        "function": {"name": "cast_float", "description": "Cast the input argument x
-        to a float.", "parameters": {"type": "object", "properties": {"x": {"title":
-        "X", "type": "string"}}, "required": ["x"]}}}, {"type": "function", "function":
-        {"name": "cast_int", "description": "Cast the input argument x to an integer.",
-        "parameters": {"type": "object", "properties": {"x": {"title": "X", "type":
-        "number"}}, "required": ["x"]}}}]}'
+        The 7 day forecast for New York is [...]"},{"role":"user","content":"Write a
+        5 word story via print_story"},{"role":"assistant","content":"Thought: . Based
+        on this reasoning, let''s select the appropriate tool!\nAction: "},{"role":"user","content":"Continue..."}],"model":"gpt-4o","n":1,"stop":["Observation:","Action:"],"temperature":0.1,"tool_choice":"required","tools":[{"type":"function","function":{"name":"print_story","description":"Print
+        a story.","parameters":{"type":"object","properties":{"story":{"description":"Story
+        to print.","title":"Story","type":"string"}},"required":["story"]}}},{"type":"function","function":{"name":"cast_float","description":"Cast
+        the input argument x to a float.","parameters":{"type":"object","properties":{"x":{"title":"X","type":"string"}},"required":["x"]}}},{"type":"function","function":{"name":"cast_int","description":"Cast
+        the input argument x to an integer.","parameters":{"type":"object","properties":{"x":{"title":"X","type":"number"}},"required":["x"]}}}]}'
       headers:
         accept:
           - application/json
@@ -144,13 +129,13 @@ interactions:
         connection:
           - keep-alive
         content-length:
-          - "1641"
+          - "1503"
         content-type:
           - application/json
         host:
           - api.openai.com
         user-agent:
-          - AsyncOpenAI/Python 1.54.3
+          - AsyncOpenAI/Python 1.56.2
         x-stainless-arch:
           - arm64
         x-stainless-async:
@@ -160,7 +145,7 @@ interactions:
         x-stainless-os:
           - MacOS
         x-stainless-package-version:
-          - 1.54.3
+          - 1.56.2
         x-stainless-raw-response:
           - "true"
         x-stainless-retry-count:
@@ -174,20 +159,20 @@ interactions:
     response:
       body:
         string: !!binary |
-          H4sIAAAAAAAAA4xTTY/TMBC951dYc25XSb83t+UALCCE6FYrlqLI60wTg2Nb9gSoqv53FKdN0rJI
-          5GBZ8+a9mXmeHCLGQOaQMhAlJ1FZNb77vM7jvZh/eHg3fZCvna0XP83m0+39ev20h1HDMM/fUdCZ
-          dSNMZRWSNLqFhUNO2Kgmy2mSxKvJYhGAyuSoGlphaTwz40k8mY3j1ThenIilkQI9pOxrxBhjh3A2
-          Leocf0PK4tE5UqH3vEBIuyTGwBnVRIB7Lz1xTTDqQWE0oW661rVSA4CMUZngSvWF2+8wuPc+caWy
-          x/n7DdZvnubrV2b39k59XE4fN/dfkkG9VnpvQ0O7WovOnwHexdOrYoyB5lXgWic1ZZ6M21/RGQPu
-          irpCTU3rcNhCSNtCuoV1rT0S+1VKb9F55lE4JM/IMC2Lkm62cIQLuWP00v3bwCmHu9pzdbLwFD92
-          b6JMYZ159lcWw05q6cvMIfdh1KHj0blaqAP1xaOCdaaylJH5gbqRncyXrSr0OzdAT+sBZIirQXx5
-          Zl3oZTkSl+HVu0UTXJSY99R+4XidSzMAosHsf3fzknY7v9TF/8j3gBBoCfPMOsyluJy4T3PY/JL/
-          SutcDg2D33vCKttJXaAL+xWW1GbJ/DZfTWeJEBAdoz8AAAD//wMA8IZlSh4EAAA=
+          H4sIAAAAAAAAAwAAAP//jFPBbtswDL37KwSek8JL0jj1rdiAnVpgyYBhWApDkRlbnSxpEr0mDfLv
+          g+XEdrIWqA+CwMf3SD7Rh4gxkDmkDETJSVRWje/zeFl+/V4k/svt9pvQO4MP8z8/fibJ5+UURg3D
+          bJ5R0Jl1I0xlFZI0uoWFQ07YqH5KptPF7d1ifheAyuSoGlphaTwz40k8mY3jxTien4ilkQI9pOxX
+          xBhjh3A2Leocd5CyeHSOVOg9LxDSLokxcEY1EeDeS09cE4x6UBhNqJuuda3UACBjVCa4Un3h9jsM
+          7r1PXKlsVeySh+dXuv9bvarlZEnx46NdvcwG9VrpvQ0NbWstOn8GeBdPr4oxBppXgWud1JR5Mm5/
+          RWcMuCvqCjU1rcNhDSFtDekaVrX2SOyllN6i88yjcEiekWFaFiXdrOEIF3LH6K3708Aph9vac3Wy
+          8BQ/dm+iTGGd2fgri2ErtfRl5pD7MOrQ8ehcLdSB+uJRwTpTWcrI/EbdyE5mk1YV+p0boKf1ADLE
+          1SA+P7Mu9LIcicvw6t2iCS5KzHtqv3C8zqUZANFg9v+7eUu7nV/q4iPyPSAEWsI8sw5zKS4n7tMc
+          Nr/ke2mdy6Fh8HtPWGVbqQt0Yb/CktpMJILHuOGCQ3SM/gEAAP//AwBbSysBHgQAAA==
       headers:
         CF-Cache-Status:
           - DYNAMIC
         CF-RAY:
-          - 8df951068d23ce5c-SJC
+          - 8effbad959279e56-SJC
         Connection:
           - keep-alive
         Content-Encoding:
@@ -195,7 +180,7 @@ interactions:
         Content-Type:
           - application/json
         Date:
-          - Fri, 08 Nov 2024 23:24:26 GMT
+          - Tue, 10 Dec 2024 19:44:30 GMT
         Server:
           - cloudflare
         Transfer-Encoding:
@@ -209,7 +194,7 @@ interactions:
         openai-organization:
           - future-house-xr4tdh
         openai-processing-ms:
-          - "413"
+          - "560"
         openai-version:
           - "2020-10-01"
         strict-transport-security:
@@ -221,13 +206,13 @@ interactions:
         x-ratelimit-remaining-requests:
           - "9999"
         x-ratelimit-remaining-tokens:
-          - "29999812"
+          - "29999828"
         x-ratelimit-reset-requests:
           - 6ms
         x-ratelimit-reset-tokens:
           - 0s
         x-request-id:
-          - req_1a8816a1d18402ce51ea115d4b0ef4e8
+          - req_c68421e4837ba73aad651c4790c88a0d
       status:
         code: 200
         message: OK
diff --git a/tests/cassettes/TestReActAgent.test_react_dummyenv[True-gpt-4-turbo].yaml b/tests/cassettes/TestReActAgent.test_react_dummyenv[True-gpt-4-turbo].yaml
index e031caa5..b4bb04fa 100644
--- a/tests/cassettes/TestReActAgent.test_react_dummyenv[True-gpt-4-turbo].yaml
+++ b/tests/cassettes/TestReActAgent.test_react_dummyenv[True-gpt-4-turbo].yaml
@@ -1,7 +1,7 @@
 interactions:
   - request:
       body:
-        '{"messages": [{"role": "system", "content": "Answer the following questions
+        '{"messages":[{"role":"system","content":"Answer the following questions
         as best you can. You have access to the following tools:\n\nNAME: print_story\n\nSYNOPSIS:\n    print_story(string
         story)\n\nDESCRIPTION:\n    Print a story.\n\nPARAMETERS:\n    story (string):
         Story to print.\n\nNAME: cast_float\n\nSYNOPSIS:\n    cast_float(string x)\n\nDESCRIPTION:\n    Cast
@@ -14,9 +14,8 @@ interactions:
         tuple\nObservation: the result of the action\n... (this Thought/Action/Action
         Input/Observation can repeat N times)\n\nExample:\n\nThought: I need to use
         the get_weather tool\nAction: get_weather\nAction Input: \"New York\", 7\nObservation:
-        The 7 day forecast for New York is [...]"}, {"role": "user", "content": "Write
-        a 5 word story via print_story"}], "model": "gpt-4-turbo", "stop": ["Observation:"],
-        "temperature": 0.1}'
+        The 7 day forecast for New York is [...]"},{"role":"user","content":"Write a
+        5 word story via print_story"}],"model":"gpt-4-turbo","n":1,"stop":["Observation:"],"temperature":0.1}'
       headers:
         accept:
           - application/json
@@ -25,13 +24,13 @@ interactions:
         connection:
           - keep-alive
         content-length:
-          - "1289"
+          - "1281"
         content-type:
           - application/json
         host:
           - api.openai.com
         user-agent:
-          - AsyncOpenAI/Python 1.54.3
+          - AsyncOpenAI/Python 1.56.2
         x-stainless-arch:
           - arm64
         x-stainless-async:
@@ -41,7 +40,7 @@ interactions:
         x-stainless-os:
           - MacOS
         x-stainless-package-version:
-          - 1.54.3
+          - 1.56.2
         x-stainless-raw-response:
           - "true"
         x-stainless-retry-count:
@@ -55,20 +54,20 @@ interactions:
     response:
       body:
         string: !!binary |
-          H4sIAAAAAAAAA4xSTa/TMBC851esfAKpqdq+kra5VeLygAN6lBNFkWtvEtPEa+zN41VP/e/I6Uda
-          ARIXH3Z2RjOzfk0AhNEiB6Fqyap1Tbp++qIOmdrQJvz88PX5/eeneqGyj+tqTYu9GEUG7X6g4gtr
-          rKh1DbIhe4KVR8kYVaeLh+l0spxl8x5oSWMTaZXjdJ5y53eUziazeTqZp5PVmV2TURhEDt8SAIDX
-          /o0+rcYXkcNkdJm0GIKsUOTXJQDhqYkTIUMwgaVlMRpARZbR9tY3NXVVzTk8gkXUwAQn4yBh5w2W
-          EJj8ARTZKGRsBVQCvkjFzQFK84zwi7wO461dqxg+B+eN5aKnXYbwaF3HObzZik8UGPZ4CFBSZzV0
-          VqOHQKUcb8Xbrb316bHsgoxd2a5pzvPjNXhDlfO0C2f8Oi+NNaEuPMpANoYMTE706DEB+N4X3N11
-          Jpyn1nHBtEcbBWfZ4qQnhrsO6MO5fcHEsrlhrS6sO71CI0vThJsTCSVVjXqgDveUnTZ0AyQ3qf90
-          8zftU3Jjq/+RHwCl0DHqwnnURt0nHtY8xm//r7Vry71hEQ6BsS1KYyv0/ceI9yhdod/p5TxbrjIU
-          yTH5DQAA//8DAIt6yEiCAwAA
+          H4sIAAAAAAAAAwAAAP//jFJdi9swEHz3r1j01EIc3CTNXfyWUsodHH0ohUKbYmRpbevO1uqk9TXp
+          cf/9kPPhhLbQFz3s7Awzs3pOAITRIgehGsmqc2261tkXpNWnx8dwX/3++KFbW9x+/v7tZntz9yQm
+          kUHlPSo+sqaKOtciG7J7WHmUjFH13dV8fv1+db1cDkBHGttIqx2ni5R7X1I6y2aLNFuk2erAbsgo
+          DCKHHwkAwPPwRp9W41bkkE2Okw5DkDWK/LQEIDy1cSJkCCawtCwmI6jIMtrB+teG+rrhHG7BImpg
+          gr1xkFB6gxUEJr8DRTYKGVsDVYBbqbjdQWWeEH6R12G6sWsVw+fgvLFcDLTjEG6t6zmHNxtxR4Hh
+          AXcBKuqtht5q9BCoktONeLux5z49Vn2QsSvbt+1h/nIK3lLtPJXhgJ/mlbEmNIVHGcjGkIHJiQF9
+          SQB+DgX3F50J56lzXDA9oI2Cs+XVXk+Mdx3R+aF9wcSyPWOtjqwLvUIjS9OGsxMJJVWDeqSO95S9
+          NnQGJGep/3TzN+19cmPr/5EfAaXQMerCedRGXSYe1zzGb/+vtVPLg2ERdoGxKypja/TDx4j3qFyR
+          VctshmWppUheklcAAAD//wMAcM+lToIDAAA=
       headers:
         CF-Cache-Status:
           - DYNAMIC
         CF-RAY:
-          - 8df950f96961253c-SJC
+          - 8effbac5ebcf6807-SJC
         Connection:
           - keep-alive
         Content-Encoding:
@@ -76,14 +75,14 @@ interactions:
         Content-Type:
           - application/json
         Date:
-          - Fri, 08 Nov 2024 23:24:24 GMT
+          - Tue, 10 Dec 2024 19:44:28 GMT
         Server:
           - cloudflare
         Set-Cookie:
-          - __cf_bm=zHuQMrOnwgyJc9MJ4XFrqssgthlTjUOSHpl0GoPmrwQ-1731108264-1.0.1.1-8_IgZ7QNlTyIRRQKL02MJJiYMzDbXtU8Pcl.bpULr32wFLZAG0DgBW2mUXVsh4Y7Lbysh33MgfiQwnM.YfmIfQ;
-            path=/; expires=Fri, 08-Nov-24 23:54:24 GMT; domain=.api.openai.com; HttpOnly;
+          - __cf_bm=ag5bmY5z6vTF_IKcj4m7ElEJQwLmTYjziW1tNojZH9U-1733859868-1.0.1.1-9GB6fsuRS4E08Ss199SiOE..koXUywi4OS4mOD0zj6m5y1e1s4BUGY5DOQxg3YQy9SyvDvo9I7_squiXUt0tZw;
+            path=/; expires=Tue, 10-Dec-24 20:14:28 GMT; domain=.api.openai.com; HttpOnly;
             Secure; SameSite=None
-          - _cfuvid=JT28sKsW8c619.li_Qj8b5CEv0QVhb4OxfDhw2Aft8Q-1731108264989-0.0.1.1-604800000;
+          - _cfuvid=biQ.Z.yDV53jhZ6_fSkBuJv_RcF3Ie8_kw3D3Nb81WU-1733859868146-0.0.1.1-604800000;
             path=/; domain=.api.openai.com; HttpOnly; Secure; SameSite=None
         Transfer-Encoding:
           - chunked
@@ -96,25 +95,25 @@ interactions:
         openai-organization:
           - future-house-xr4tdh
         openai-processing-ms:
-          - "964"
+          - "1450"
         openai-version:
           - "2020-10-01"
         strict-transport-security:
           - max-age=31536000; includeSubDomains; preload
         x-ratelimit-limit-requests:
-          - "3000"
+          - "10000"
         x-ratelimit-limit-tokens:
-          - "250000"
+          - "2000000"
         x-ratelimit-remaining-requests:
-          - "2998"
+          - "9999"
         x-ratelimit-remaining-tokens:
-          - "249472"
+          - "1999710"
         x-ratelimit-reset-requests:
-          - 27ms
+          - 6ms
         x-ratelimit-reset-tokens:
-          - 126ms
+          - 8ms
         x-request-id:
-          - req_bd4575cd1b4a7b53a973861d90b72214
+          - req_db2829aa97ebbcfcfa11b8136e71e575
       status:
         code: 200
         message: OK
diff --git a/tests/cassettes/TestSimpleAgent.test_agent_grad[gpt-4o-mini-2024-07-18].yaml b/tests/cassettes/TestSimpleAgent.test_agent_grad[gpt-4o-mini-2024-07-18].yaml
index 10620d4c..cc1d9015 100644
--- a/tests/cassettes/TestSimpleAgent.test_agent_grad[gpt-4o-mini-2024-07-18].yaml
+++ b/tests/cassettes/TestSimpleAgent.test_agent_grad[gpt-4o-mini-2024-07-18].yaml
@@ -1,17 +1,11 @@
 interactions:
   - request:
       body:
-        '{"messages": [{"role": "user", "content": "Write a 5 word story via print_story"}],
-        "model": "gpt-4o-mini-2024-07-18", "temperature": 0.1, "tool_choice": "required",
-        "tools": [{"type": "function", "function": {"name": "print_story", "description":
-        "Print a story.", "parameters": {"type": "object", "properties": {"story": {"description":
-        "Story to print.", "title": "Story", "type": "string"}}, "required": ["story"]}}},
-        {"type": "function", "function": {"name": "cast_float", "description": "Cast
-        the input argument x to a float.", "parameters": {"type": "object", "properties":
-        {"x": {"title": "X", "type": "string"}}, "required": ["x"]}}}, {"type": "function",
-        "function": {"name": "cast_int", "description": "Cast the input argument x to
-        an integer.", "parameters": {"type": "object", "properties": {"x": {"title":
-        "X", "type": "number"}}, "required": ["x"]}}}]}'
+        '{"messages":[{"role":"user","content":"Write a 5 word story via print_story"}],"model":"gpt-4o-mini-2024-07-18","n":1,"temperature":0.1,"tool_choice":"required","tools":[{"type":"function","function":{"name":"print_story","description":"Print
+        a story.","parameters":{"type":"object","properties":{"story":{"description":"Story
+        to print.","title":"Story","type":"string"}},"required":["story"]}}},{"type":"function","function":{"name":"cast_float","description":"Cast
+        the input argument x to a float.","parameters":{"type":"object","properties":{"x":{"title":"X","type":"string"}},"required":["x"]}}},{"type":"function","function":{"name":"cast_int","description":"Cast
+        the input argument x to an integer.","parameters":{"type":"object","properties":{"x":{"title":"X","type":"number"}},"required":["x"]}}}]}'
       headers:
         accept:
           - application/json
@@ -20,13 +14,13 @@ interactions:
         connection:
           - keep-alive
         content-length:
-          - "867"
+          - "806"
         content-type:
           - application/json
         host:
           - api.openai.com
         user-agent:
-          - AsyncOpenAI/Python 1.54.3
+          - AsyncOpenAI/Python 1.56.2
         x-stainless-arch:
           - arm64
         x-stainless-async:
@@ -36,11 +30,11 @@ interactions:
         x-stainless-os:
           - MacOS
         x-stainless-package-version:
-          - 1.54.3
+          - 1.56.2
         x-stainless-raw-response:
           - "true"
         x-stainless-retry-count:
-          - "1"
+          - "0"
         x-stainless-runtime:
           - CPython
         x-stainless-runtime-version:
@@ -50,20 +44,20 @@ interactions:
     response:
       body:
         string: !!binary |
-          H4sIAAAAAAAAA4xTTW+bQBC98ytWczYREKcm3HyJqqqtrNSt+uEKbXYHvPWyQ3aXKK7l/16BbcBu
-          IoUDQvPmvZl5M+wCxkBJyBiINfeiqnU4v/8ik7lO7pZfo01FOk3WT/Pm7+3i52IZwaRl0MMfFP7E
-          uhJU1Rq9InOAhUXusVWNZ9dxHKXJu7QDKpKoW1pZ+3BKYaWMCpMomYbRLIzTI3tNSqCDjP0KGGNs
-          173bPo3EZ8hYNDlFKnSOlwhZn8QYWNJtBLhzynluPEwGUJDxaNrWTaP1CPBEOhdc66Hw4dmNvgez
-          uNb5N/8j/vBYLW6Kz+/v8dP27jld0uP0+6jeQXpbdw0VjRG9SSO8j2cXxRgDw6uOW1tlfO482e0F
-          nTHgtmwqNL5tHXYr6NJWkK3gIznPNridMElkWUEWn9AyTWKD8moFezjT2gcvff8e2WSxaBzXR/+O
-          8X2/EE1lbenBXfgLhTLKrXOL3HVzju0OTtW6OtCcbRRqS1Xtc08bNK1sHB/XD8PVDWhyAj15rkes
-          6xNwppdL9Fx1K++vTHCxRjlQh2vjjVQ0AoLR7P9385L2YX5lyrfID4AQWHuUeW1RKnE+8ZBmsf0p
-          X0vrXe4aBrd1Hqu8UKZE2x1Xd6F1Xtzc8jSeJrMCgn3wDwAA//8DAH4X1SggBAAA
+          H4sIAAAAAAAAAwAAAP//jFNNj9MwEL3nV1hzblb9WHZDbgg4IAGLkBbKUhQ59jR11/FYtgMbVf3v
+          q7htkpZFIocomjfvzcybyS5hDJSEnIHY8CBqq9M3cvoVl7M7H9Dq7cN3zO5U+eOdfpLYVjDpGFRu
+          UYQT60pQbTUGReYAC4c8YKc6u10sslevs5ubCNQkUXe0yob0mtJaGZXOp/PrdHqbzrIje0NKoIec
+          /UwYY2wX312fRuIT5Gw6OUVq9J5XCHmfxBg40l0EuPfKB24CTAZQkAloutZNo/UICES6EFzrofDh
+          2Y2+B7O41oX4zO+3+v2fpW+X3x4+fPmU+fvGibejegfp1saG1o0RvUkjvI/nF8UYA8PryLVOmVD4
+          QK69oDMG3FVNjSZ0rcNuBTFtBfkKPpIP7BHbCZNEjq3J4W90TGjyKK9WsIczrX3y0vevkU0O143n
+          +ujfMb7vF6Kpso5Kf+EvrJVRflM45D7OObY7OVWLdaA52yhYR7UNRaBHNJ3sbHZcPwxXN6DzExgo
+          cD1iLU7AmV4hMXAVV95fmeBig3KgDtfGG6loBCSj2f/u5iXtw/zKVP8jPwBCoA0oC+tQKnE+8ZDm
+          sPsp/5XWuxwbBt/6gHWxVqZCF48rXqgtypIvRIa30xKSffIMAAD//wMA+2fWYyAEAAA=
       headers:
         CF-Cache-Status:
           - DYNAMIC
         CF-RAY:
-          - 8df951176dbe2393-SJC
+          - 8effbac5ffef06ad-SJC
         Connection:
           - keep-alive
         Content-Encoding:
@@ -71,9 +65,15 @@ interactions:
         Content-Type:
           - application/json
         Date:
-          - Fri, 08 Nov 2024 23:24:29 GMT
+          - Tue, 10 Dec 2024 19:44:27 GMT
         Server:
           - cloudflare
+        Set-Cookie:
+          - __cf_bm=C3dbW6zJBRY8Ys1xbYeud2RLlyCwWrInFVnA9iWonSk-1733859867-1.0.1.1-5lhE.LwvBpR3mP40cazViwFSprKDhNS5YV_RCYO89CtNv.WGTlUqQBdOduKwDRmvqh3tk8qwZb5NjWIKagwiiw;
+            path=/; expires=Tue, 10-Dec-24 20:14:27 GMT; domain=.api.openai.com; HttpOnly;
+            Secure; SameSite=None
+          - _cfuvid=B3mlsyUvxiiy3LKopH3ULUQ1oGGbfbGZe6GTMgcgc5A-1733859867079-0.0.1.1-604800000;
+            path=/; domain=.api.openai.com; HttpOnly; Secure; SameSite=None
         Transfer-Encoding:
           - chunked
         X-Content-Type-Options:
@@ -85,7 +85,7 @@ interactions:
         openai-organization:
           - future-house-xr4tdh
         openai-processing-ms:
-          - "527"
+          - "423"
         openai-version:
           - "2020-10-01"
         strict-transport-security:
@@ -97,13 +97,13 @@ interactions:
         x-ratelimit-remaining-requests:
           - "29999"
         x-ratelimit-remaining-tokens:
-          - "149999974"
+          - "149999973"
         x-ratelimit-reset-requests:
           - 2ms
         x-ratelimit-reset-tokens:
           - 0s
         x-request-id:
-          - req_9794c5d3946c0779e2ae985662592f7c
+          - req_97968b8e2e2867ddebbdd0b6f8ee79ee
       status:
         code: 200
         message: OK
diff --git a/tests/conftest.py b/tests/conftest.py
index ecb934f6..60ae7f4e 100644
--- a/tests/conftest.py
+++ b/tests/conftest.py
@@ -6,8 +6,7 @@
 import pytest
 import torch
 from aviary.core import DummyEnv
-
-from ldp.utils import configure_log_levels
+from llmclient import configure_llm_logs
 
 from . import CASSETTES_DIR
 
@@ -21,7 +20,7 @@ def fixture_dummy_env() -> DummyEnv:
 
 @pytest.fixture(scope="session", autouse=True)
 def _fixture_set_up_environment() -> None:
-    configure_log_levels()
+    configure_llm_logs()
 
 
 def set_seed(seed: int | None) -> None:
diff --git a/tests/test_agents.py b/tests/test_agents.py
index 0c34a2cd..1a488607 100644
--- a/tests/test_agents.py
+++ b/tests/test_agents.py
@@ -11,6 +11,7 @@
 from aviary.core import DummyEnv, Message, Tool, ToolCall, ToolRequestMessage
 from aviary.message import EnvStateMessage
 from httpx import ASGITransport, AsyncClient
+from llmclient import MultipleCompletionLLMModel as LLMModel
 from pydantic import BaseModel, Field
 
 from ldp.agent import (
@@ -34,7 +35,6 @@
     ReActModuleSinglePrompt,
     ToolDescriptionMethods,
 )
-from ldp.llms import LLMModel
 
 from . import CILLMModelNames
 from .conftest import IN_GITHUB_ACTIONS, VCR_DEFAULT_MATCH_ON
diff --git a/tests/test_embeddings.py b/tests/test_embeddings.py
deleted file mode 100644
index 514a846f..00000000
--- a/tests/test_embeddings.py
+++ /dev/null
@@ -1,139 +0,0 @@
-import asyncio
-from unittest.mock import MagicMock, patch
-
-import litellm
-import numpy as np
-import pytest
-from litellm.caching import Cache, InMemoryCache
-from pytest_subtests import SubTests
-
-from ldp.llms import (
-    EmbeddingModel,
-    HybridEmbeddingModel,
-    LiteEmbeddingModel,
-    SparseEmbeddingModel,
-)
-
-
-class TestLiteEmbeddingModel:
-    @pytest.mark.asyncio
-    async def test_embed_texts(self) -> None:
-        texts = ["Hello", "World"]
-        batch_size = 1  # NOTE: this affects the mock below
-        model = LiteEmbeddingModel(name="stub", batch_size=1)
-        with patch(
-            "litellm.aembedding",
-            autospec=True,
-            side_effect=[
-                MagicMock(data=[{"embedding": [1.0, 2.0]}]),
-                MagicMock(data=[{"embedding": [3.0, 4.0]}]),
-            ],
-        ) as mock_aembedding:
-            embeddings = await model.embed_texts(texts)
-
-        assert np.allclose(embeddings[0], [1.0, 2.0])
-        assert np.allclose(embeddings[1], [3.0, 4.0])
-        assert mock_aembedding.call_count == len(texts) / batch_size
-
-    @pytest.mark.parametrize(
-        ("model_name", "expected_dimensions"),
-        [
-            ("stub", None),
-            ("text-embedding-ada-002", 1536),
-            ("text-embedding-3-small", 1536),
-        ],
-    )
-    def test_model_dimension_inference(
-        self, model_name: str, expected_dimensions: int | None
-    ) -> None:
-        assert LiteEmbeddingModel(name=model_name).dimensions == expected_dimensions
-
-    @pytest.mark.asyncio
-    async def test_can_change_dimension(self) -> None:
-        """We run this one for real, because want to test end to end."""
-        stub_texts = ["test1", "test2"]
-
-        model = LiteEmbeddingModel(name="text-embedding-3-small")
-        assert model.dimensions == 1536
-
-        model = LiteEmbeddingModel(name="text-embedding-3-small", dimensions=8)
-        assert model.dimensions == 8
-        etext1, etext2 = await model.embed_texts(stub_texts)
-        assert len(etext1) == len(etext2) == 8
-
-    @pytest.mark.vcr
-    @pytest.mark.asyncio
-    async def test_caching(self) -> None:
-        model = LiteEmbeddingModel(
-            name="text-embedding-3-small", dimensions=8, embed_kwargs={"caching": True}
-        )
-        # Make sure there is no existing cache.
-        with patch("litellm.cache", None):
-            # now create a new cache
-            litellm.cache = Cache()
-            assert isinstance(litellm.cache.cache, InMemoryCache)
-            assert len(litellm.cache.cache.cache_dict) == 0
-
-            _ = await model.embed_texts(["test1"])
-            # need to do this to see the data propagated to cache
-            await asyncio.sleep(0.0)
-
-            # Check the cache entry was made
-            assert len(litellm.cache.cache.cache_dict) == 1
-
-
-@pytest.mark.asyncio
-async def test_sparse_embedding_model(subtests: SubTests):
-    with subtests.test("1D sparse"):
-        ndim = 1
-        expected_output = [[1.0], [1.0]]
-
-        model = SparseEmbeddingModel(dimensions=ndim)
-        result = await model.embed_texts(["test1", "test2"])
-
-        assert result == expected_output
-
-    with subtests.test("large sparse"):
-        ndim = 1024
-
-        model = SparseEmbeddingModel(dimensions=ndim)
-        result = await model.embed_texts(["hello test", "go hello"])
-
-        assert max(result[0]) == max(result[1]) == 0.5
-
-    with subtests.test("default sparse"):
-        model = SparseEmbeddingModel()
-        result = await model.embed_texts(["test1 hello", "test2 hello"])
-
-        assert pytest.approx(sum(result[0]), abs=1e-6) == pytest.approx(
-            sum(result[1]), abs=1e-6
-        )
-
-
-@pytest.mark.asyncio
-async def test_hybrid_embedding_model() -> None:
-    hybrid_model = HybridEmbeddingModel(
-        models=[LiteEmbeddingModel(), SparseEmbeddingModel()]
-    )
-
-    # Mock the embedded documents of Lite and Sparse models
-    with (
-        patch.object(LiteEmbeddingModel, "embed_texts", return_value=[[1.0], [2.0]]),
-        patch.object(SparseEmbeddingModel, "embed_texts", return_value=[[3.0], [4.0]]),
-    ):
-        result = await hybrid_model.embed_texts(["hello", "world"])
-    assert result.tolist() == [[1.0, 3.0], [2.0, 4.0]]
-
-
-@pytest.mark.asyncio
-async def test_class_constructor() -> None:
-    original_name = "hybrid-text-embedding-3-small"
-    model = EmbeddingModel.from_name(original_name)
-    assert isinstance(model, HybridEmbeddingModel)
-    assert model.name == original_name
-    dense_model, sparse_model = model.models
-    assert dense_model.name == "text-embedding-3-small"
-    assert dense_model.dimensions == 1536
-    assert sparse_model.name == "sparse"
-    assert sparse_model.dimensions == 256
-    assert model.dimensions == 1792
diff --git a/tests/test_llms.py b/tests/test_llms.py
deleted file mode 100644
index f9037718..00000000
--- a/tests/test_llms.py
+++ /dev/null
@@ -1,356 +0,0 @@
-from typing import Any, ClassVar
-from unittest.mock import Mock
-
-import litellm
-import numpy as np
-import pytest
-from aviary.core import DummyEnv, Message, Tool, ToolRequestMessage
-from pydantic import BaseModel, Field
-
-from ldp.llms import (
-    JSONSchemaValidationError,
-    LLMModel,
-    LLMResult,
-    MultipleCompletionLLMModel,
-    validate_json_completion,
-)
-
-from . import CILLMModelNames
-
-
-def test_json_schema_validation() -> None:
-    # Invalid JSON
-    mock_completion1 = Mock()
-    mock_completion1.choices = [Mock()]
-    mock_completion1.choices[0].message.content = "not a json"
-    # Invalid schema
-    mock_completion2 = Mock()
-    mock_completion2.choices = [Mock()]
-    mock_completion2.choices[0].message.content = '{"name": "John", "age": "nan"}'
-    # Valid schema
-    mock_completion3 = Mock()
-    mock_completion3.choices = [Mock()]
-    mock_completion3.choices[0].message.content = '{"name": "John", "age": 30}'
-
-    class DummyModel(BaseModel):
-        name: str
-        age: int
-
-    with pytest.raises(JSONSchemaValidationError):
-        validate_json_completion(mock_completion1, DummyModel)
-    with pytest.raises(JSONSchemaValidationError):
-        validate_json_completion(mock_completion2, DummyModel)
-    validate_json_completion(mock_completion3, DummyModel)
-
-
-@pytest.mark.parametrize(
-    "model_name", ["gpt-3.5-turbo", CILLMModelNames.ANTHROPIC.value]
-)
-@pytest.mark.asyncio
-async def test_achat(model_name: str) -> None:
-    model = LLMModel(name=model_name)
-    response = await model.achat(
-        messages=[
-            Message(content="What are three things I should do today?"),
-        ]
-    )
-
-    assert len(response.choices) == 1
-
-    # Check we can iterate through the response
-    async for chunk in await model.achat_iter(
-        messages=[
-            Message(content="What are three things I should do today?"),
-        ]
-    ):
-        assert len(chunk.choices) == 1
-
-
-@pytest.mark.parametrize(
-    "model_name", [CILLMModelNames.OPENAI.value, CILLMModelNames.ANTHROPIC.value]
-)
-@pytest.mark.flaky(reruns=3, only_on=[litellm.exceptions.APIConnectionError])
-@pytest.mark.asyncio
-async def test_tools(dummy_env: DummyEnv, model_name: str) -> None:
-    model = LLMModel(name=model_name)
-    messages = [
-        Message(content="What are three things I should do today?"),
-    ]
-
-    def get_todo_list(n: int) -> str:
-        """Get todo list for today.
-
-        Args:
-            n: number of items to return
-        """
-        return "\n".join(["Go for a walk", "Read a book", "Call a friend"][:n])
-
-    tool = Tool.from_function(get_todo_list)
-    dummy_env.tools = [tool]
-    result = await model.call(messages, tools=dummy_env.tools)
-    assert result.completion_count > 0
-
-    # try specifying tool choice
-    result = await model.call(messages, tools=dummy_env.tools, tool_choice=tool)
-    assert result.completion_count > 0, "Tool choice failed to execute tool"
-    assert result.messages
-    (tool_request_message,) = result.messages
-    assert isinstance(tool_request_message, ToolRequestMessage)
-
-    new_messages = await dummy_env.exec_tool_calls(tool_request_message)
-    (new_message,) = new_messages
-    assert new_message.content == "Go for a walk\nRead a book\nCall a friend"
-    assert new_message.tool_call_id == tool_request_message.tool_calls[0].id
-
-    def get_todo_list_no_args() -> str:
-        """Get todo list for today."""
-        return "\n".join(["Go for a walk", "Read a book", "Call a friend"])
-
-    tool = Tool.from_function(get_todo_list_no_args)
-    dummy_env.tools = [tool]
-    result = await model.call(messages, tools=dummy_env.tools)
-    assert result.completion_count > 0
-    assert result.messages
-    (tool_request_message,) = result.messages
-    assert isinstance(tool_request_message, ToolRequestMessage)
-
-    new_messages = await dummy_env.exec_tool_calls(tool_request_message)
-    (new_message,) = new_messages
-    assert new_message.content == "Go for a walk\nRead a book\nCall a friend"
-    assert new_message.tool_call_id == tool_request_message.tool_calls[0].id
-
-    # ok now try with multiple functions
-    messages = [
-        Message(
-            content=(
-                "What items will I have time to accomplish on my todo list today based"
-                " on my calendar?"
-            )
-        ),
-    ]
-
-    def get_calendar() -> str:
-        """Get text version of calendar for today."""
-        return "9:00am Wake-up\n10:00pm Go to bed\n"
-
-    tool2 = Tool.from_function(get_calendar)
-    dummy_env.tools = [tool, tool2]
-    result = await model.call(messages, tools=dummy_env.tools)
-    assert result.messages
-    (tool_request_message,) = result.messages
-    assert isinstance(tool_request_message, ToolRequestMessage)
-    new_messages = await dummy_env.exec_tool_calls(tool_request_message)
-    if model_name.startswith("claude"):
-        # Anthropic not always so smart
-        assert 1 <= len(new_messages) <= 2
-    else:
-        assert len(new_messages) == 2
-
-    # ok now try continuation - I AM NOT SURE IF THIS IS VALID?
-    # TODO: - supported on openai, but not litellm
-    # messages = messages + result.messages + new_messages
-    # result = await model.call(messages)
-
-
-class DummyOutputSchema(BaseModel):
-    name: str
-    age: int
-
-
-class TestMultipleCompletionLLMModel:
-    NUM_COMPLETIONS: ClassVar[int] = 2
-    DEFAULT_CONFIG: ClassVar[dict] = {"n": NUM_COMPLETIONS}
-    MODEL_CLS: ClassVar[type[MultipleCompletionLLMModel]] = MultipleCompletionLLMModel
-
-    async def call_model(
-        self, model: MultipleCompletionLLMModel, *args, **kwargs
-    ) -> list[LLMResult]:
-        return await model.call(*args, **kwargs)
-
-    @pytest.mark.vcr
-    @pytest.mark.parametrize("model_name", ["gpt-3.5-turbo"])
-    @pytest.mark.asyncio
-    async def test_model(self, model_name: str) -> None:
-        # Make model_name an arg so that TestLLMModel can parametrize it
-        # only testing OpenAI, as other APIs don't support n>1
-        model = self.MODEL_CLS(name=model_name, config=self.DEFAULT_CONFIG)
-        messages = [
-            Message(role="system", content="Respond with single words."),
-            Message(content="Hello, how are you?"),
-        ]
-        results = await self.call_model(model, messages)
-        assert len(results) == self.NUM_COMPLETIONS
-
-        for result in results:
-            assert result.prompt_count > 0
-            assert result.completion_count > 0
-            prompt_cost, completion_cost = result.prompt_and_completion_costs
-            assert prompt_cost > 0
-            assert completion_cost > 0
-            assert result.logprob is None or result.logprob <= 0
-
-    @pytest.mark.parametrize(
-        "model_name", [CILLMModelNames.ANTHROPIC.value, "gpt-3.5-turbo"]
-    )
-    @pytest.mark.asyncio
-    async def test_streaming(self, model_name: str) -> None:
-        model = self.MODEL_CLS(name=model_name, config=self.DEFAULT_CONFIG)
-        messages = [
-            Message(role="system", content="Respond with single words."),
-            Message(content="Hello, how are you?"),
-        ]
-
-        def callback(_) -> None:
-            return
-
-        with pytest.raises(
-            NotImplementedError,
-            match="Multiple completions with callbacks is not supported",
-        ):
-            await self.call_model(model, messages, [callback])
-
-    @pytest.mark.vcr
-    @pytest.mark.asyncio
-    async def test_parameterizing_tool_from_arg_union(self) -> None:
-        def play(move: int | None) -> None:
-            """Play one turn by choosing a move.
-
-            Args:
-                move: Choose an integer to lose, choose None to win.
-            """
-
-        results = await self.call_model(
-            self.MODEL_CLS(name="gpt-3.5-turbo", config=self.DEFAULT_CONFIG),
-            messages=[Message(content="Please win.")],
-            tools=[Tool.from_function(play)],
-        )
-        assert len(results) == self.NUM_COMPLETIONS
-        for result in results:
-            assert result.messages
-            assert len(result.messages) == 1
-            assert isinstance(result.messages[0], ToolRequestMessage)
-            assert result.messages[0].tool_calls
-            assert result.messages[0].tool_calls[0].function.arguments["move"] is None
-
-    @pytest.mark.asyncio
-    @pytest.mark.vcr
-    @pytest.mark.parametrize(
-        ("model_name", "output_type"),
-        [
-            pytest.param("gpt-3.5-turbo", DummyOutputSchema, id="json-mode"),
-            pytest.param(
-                "gpt-4o", DummyOutputSchema.model_json_schema(), id="structured-outputs"
-            ),
-        ],
-    )
-    async def test_output_schema(
-        self, model_name: str, output_type: type[BaseModel] | dict[str, Any]
-    ) -> None:
-        model = self.MODEL_CLS(name=model_name, config=self.DEFAULT_CONFIG)
-        messages = [
-            Message(
-                content=(
-                    "My name is Claude and I am 1 year old. What is my name and age?"
-                )
-            ),
-        ]
-        results = await self.call_model(model, messages, output_type=output_type)
-        assert len(results) == self.NUM_COMPLETIONS
-        for result in results:
-            assert result.messages
-            assert len(result.messages) == 1
-            assert result.messages[0].content
-            DummyOutputSchema.model_validate_json(result.messages[0].content)
-
-    @pytest.mark.parametrize("model_name", [CILLMModelNames.OPENAI.value])
-    @pytest.mark.asyncio
-    @pytest.mark.vcr
-    async def test_text_image_message(self, model_name: str) -> None:
-        model = self.MODEL_CLS(name=model_name, config=self.DEFAULT_CONFIG)
-
-        # An RGB image of a red square
-        image = np.zeros((32, 32, 3), dtype=np.uint8)
-        image[:] = [255, 0, 0]  # (255 red, 0 green, 0 blue) is maximum red in RGB
-
-        results = await self.call_model(
-            model,
-            messages=[
-                Message.create_message(
-                    text="What color is this square? Respond only with the color name.",
-                    image=image,
-                )
-            ],
-        )
-        assert len(results) == self.NUM_COMPLETIONS
-        for result in results:
-            assert result.messages is not None, (
-                "Expected messages in result, but got None"
-            )
-            assert result.messages[-1].content is not None, (
-                "Expected content in message, but got None"
-            )
-            assert "red" in result.messages[-1].content.lower()
-
-
-class TestLLMModel(TestMultipleCompletionLLMModel):
-    NUM_COMPLETIONS: ClassVar[int] = 1
-    DEFAULT_CONFIG: ClassVar[dict] = {}
-    MODEL_CLS: ClassVar[type[MultipleCompletionLLMModel]] = LLMModel
-
-    async def call_model(self, model: LLMModel, *args, **kwargs) -> list[LLMResult]:  # type: ignore[override]
-        return [await model.call(*args, **kwargs)]
-
-    @pytest.mark.parametrize(
-        "model_name", [CILLMModelNames.ANTHROPIC.value, "gpt-3.5-turbo"]
-    )
-    @pytest.mark.asyncio
-    @pytest.mark.vcr
-    async def test_model(self, model_name: str) -> None:
-        await super().test_model(model_name)
-
-    @pytest.mark.vcr
-    @pytest.mark.parametrize(
-        "model_name", [CILLMModelNames.ANTHROPIC.value, "gpt-3.5-turbo"]
-    )
-    @pytest.mark.asyncio
-    async def test_streaming(self, model_name: str) -> None:
-        model = LLMModel(name=model_name)
-        messages = [
-            Message(role="system", content="Respond with single words."),
-            Message(content="Hello, how are you?"),
-        ]
-        content = []
-
-        def callback(s):
-            content.append(s)
-
-        result = await model.call(messages, [callback])
-        assert result.completion_count > 0
-        assert content
-
-    @pytest.mark.vcr
-    @pytest.mark.asyncio
-    async def test_parameterizing_tool_from_arg_union(self) -> None:
-        await super().test_parameterizing_tool_from_arg_union()
-
-    @pytest.mark.vcr
-    @pytest.mark.asyncio
-    async def test_output_type_rejected_validation(self) -> None:
-        class InstructionList(BaseModel):
-            instructions: list[str] = Field(description="list of instructions")
-
-        model = self.MODEL_CLS(name=CILLMModelNames.ANTHROPIC.value)
-        with pytest.raises(litellm.BadRequestError, match="anthropic"):
-            await model.call(
-                [Message(content="What are three things I should do today?")],
-                output_type=InstructionList,
-            )
-
-    @pytest.mark.parametrize(
-        "model_name",
-        [CILLMModelNames.ANTHROPIC.value, "gpt-4-turbo", CILLMModelNames.OPENAI.value],
-    )
-    @pytest.mark.asyncio
-    @pytest.mark.vcr
-    async def test_text_image_message(self, model_name: str) -> None:
-        await super().test_text_image_message(model_name)
diff --git a/tests/test_memory.py b/tests/test_memory.py
index 5ec2d89e..8b9509f2 100644
--- a/tests/test_memory.py
+++ b/tests/test_memory.py
@@ -1,9 +1,9 @@
 import pytest
+from llmclient import EmbeddingModel
 from pytest_subtests import SubTests
 
 from ldp.graph import Memory
 from ldp.graph.memory import UIndexMemoryModel
-from ldp.llms import EmbeddingModel
 
 
 @pytest.fixture(name="sample_memory")
diff --git a/tests/test_ops.py b/tests/test_ops.py
index 42b15089..b1585d4a 100644
--- a/tests/test_ops.py
+++ b/tests/test_ops.py
@@ -9,6 +9,7 @@
 import pytest
 import tree
 from aviary.core import DummyEnv, Message, Tool, ToolRequestMessage
+from llmclient import MultipleCompletionLLMModel as LLMModel
 
 from ldp.graph import (
     CallID,
@@ -28,7 +29,7 @@
 )
 from ldp.graph.gradient_estimators import straight_through_estimator as ste
 from ldp.graph.ops import GradInType, ResultOrValue, TOutput
-from ldp.llms import LLMModel, append_to_sys
+from ldp.llms.prompts import append_to_sys
 
 
 class StatefulFxnOp(FxnOp[TOutput]):
@@ -151,9 +152,7 @@ async def generate_story() -> str:
 
         # LLMCallOp track cost using run context
         result = llm_op.ctx.get(op_result.call_id, "result")
-        prompt_cost, completion_cost = result.prompt_and_completion_costs
-        assert prompt_cost > 0
-        assert completion_cost > 0
+        assert result.cost > 0
 
         # Environment tracks its internal costs
         assert env.total_cost > 0
diff --git a/tests/test_optimizer.py b/tests/test_optimizer.py
index 3f48d456..67e1eb22 100644
--- a/tests/test_optimizer.py
+++ b/tests/test_optimizer.py
@@ -7,6 +7,7 @@
 import tenacity
 import tree
 from aviary.core import Message
+from llmclient import MultipleCompletionLLMModel as LLMModel
 from pydantic import BaseModel, Field, JsonValue
 
 from ldp.agent import Agent, MemoryAgent, ReActAgent
@@ -38,7 +39,7 @@
     straight_through_estimator as ste,
 )
 from ldp.graph.ops import GradInType
-from ldp.llms import LLMModel, append_to_sys
+from ldp.llms.prompts import append_to_sys
 from tests import CILLMModelNames
 from tests.conftest import VCR_DEFAULT_MATCH_ON
 
@@ -349,7 +350,7 @@ async def memory_factory(
                     )
                     for mem_call_id, output_call_id, d_return, metadata in example_buffer
                 ]
-                response = await memory_distiller.call(
+                response = await memory_distiller.call_single(
                     messages=[
                         Message(
                             content=LessonEntry.make_prompt(query_airesponse_dreturns)
diff --git a/uv.lock b/uv.lock
index d22fbd6f..ae486459 100644
--- a/uv.lock
+++ b/uv.lock
@@ -5,10 +5,14 @@ resolution-markers = [
     "python_full_version < '3.12' and platform_python_implementation == 'PyPy' and sys_platform != 'linux'",
     "python_full_version < '3.12' and platform_python_implementation != 'PyPy' and sys_platform == 'linux'",
     "python_full_version < '3.12' and platform_python_implementation != 'PyPy' and sys_platform != 'linux'",
-    "python_full_version >= '3.12' and platform_python_implementation == 'PyPy' and sys_platform == 'linux'",
-    "python_full_version >= '3.12' and platform_python_implementation == 'PyPy' and sys_platform != 'linux'",
-    "python_full_version >= '3.12' and platform_python_implementation != 'PyPy' and sys_platform == 'linux'",
-    "python_full_version >= '3.12' and platform_python_implementation != 'PyPy' and sys_platform != 'linux'",
+    "python_full_version == '3.12.*' and platform_python_implementation == 'PyPy' and sys_platform == 'linux'",
+    "python_full_version >= '3.13' and platform_python_implementation == 'PyPy' and sys_platform == 'linux'",
+    "python_full_version == '3.12.*' and platform_python_implementation == 'PyPy' and sys_platform != 'linux'",
+    "python_full_version >= '3.13' and platform_python_implementation == 'PyPy' and sys_platform != 'linux'",
+    "python_full_version == '3.12.*' and platform_python_implementation != 'PyPy' and sys_platform == 'linux'",
+    "python_full_version >= '3.13' and platform_python_implementation != 'PyPy' and sys_platform == 'linux'",
+    "python_full_version == '3.12.*' and platform_python_implementation != 'PyPy' and sys_platform != 'linux'",
+    "python_full_version >= '3.13' and platform_python_implementation != 'PyPy' and sys_platform != 'linux'",
 ]
 
 [[package]]
@@ -153,6 +157,15 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/25/8a/c46dcc25341b5bce5472c718902eb3d38600a903b14fa6aeecef3f21a46f/asttokens-3.0.0-py3-none-any.whl", hash = "sha256:e3078351a059199dd5138cb1c706e6430c05eff2ff136af5eb4790f9d28932e2", size = 26918 },
 ]
 
+[[package]]
+name = "async-timeout"
+version = "4.0.3"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/87/d6/21b30a550dafea84b1b8eee21b5e23fa16d010ae006011221f33dcd8d7f8/async-timeout-4.0.3.tar.gz", hash = "sha256:4640d96be84d82d02ed59ea2b7105a0f7b33abe8703703cd0ab0bf87c427522f", size = 8345 }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/a7/fa/e01228c2938de91d47b307831c62ab9e4001e747789d0b05baf779a6488c/async_timeout-4.0.3-py3-none-any.whl", hash = "sha256:7405140ff1230c310e51dc27b3145b9092d659ce68ff733fb0cefe3ee42be028", size = 5721 },
+]
+
 [[package]]
 name = "attrs"
 version = "24.2.0"
@@ -403,6 +416,33 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/c1/31/1ae946f11dfbd229222e6d6ad8e7bd1891d3d48bde5fbf7a0beb9491f8e3/contourpy-1.3.1-cp313-cp313t-win_amd64.whl", hash = "sha256:287ccc248c9e0d0566934e7d606201abd74761b5703d804ff3df8935f523d546", size = 236668 },
 ]
 
+[[package]]
+name = "coredis"
+version = "4.17.0"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "async-timeout" },
+    { name = "deprecated" },
+    { name = "packaging" },
+    { name = "pympler" },
+    { name = "typing-extensions" },
+    { name = "wrapt" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/ef/0c/0f2fb1cedd224666ef08e898447bb9cf4d1e98a86b03119f1c6513093ddc/coredis-4.17.0.tar.gz", hash = "sha256:04e9976e71a42004dfe19a862c648b4047bf813e15184cddfd3cb37eb704b83f", size = 243157 }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/c1/2c/2335e476f0c0b33eea53c307169bcafe9c19a4b277738258eb80354ee90c/coredis-4.17.0-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:f3050806b4854a6624e3c2efa013b540265d88e766f815963d447c116240d75d", size = 330690 },
+    { url = "https://files.pythonhosted.org/packages/6a/b1/3c24a708b24f8e2566b1b91b64b4dc75f74633b875def19f2ac0fa03a0a0/coredis-4.17.0-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:5f0f1044bdafc93f421e59e711da762c6c741ab76df0c12a42c447c1db1fcd75", size = 328051 },
+    { url = "https://files.pythonhosted.org/packages/0f/a6/e5a8add1ae7b31240248528f669127e5fd347c69625a9b423965a5902302/coredis-4.17.0-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:f1befa7db121978fd0995151af5d15ce5e37a14847797c3fbd9403882f21b48c", size = 352651 },
+    { url = "https://files.pythonhosted.org/packages/b8/d1/0ece1b888547ec26f4d33be30513cd44c77df25c9f943e7d3c20b49cc634/coredis-4.17.0-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:52583dcef671c8d3a1cbecbf81cd630b1a72f946cf46601016c4f85d3f12a4a1", size = 355472 },
+    { url = "https://files.pythonhosted.org/packages/00/c2/771bafa43c37d8c968804b6bb34063eb631b5d2377db31bca6d784131f48/coredis-4.17.0-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:845f5c0bb7012609a1f41f8308e5166c01f162599af33cb001bd2b0d6a4386f5", size = 358740 },
+    { url = "https://files.pythonhosted.org/packages/fb/d3/90846efc003d692c46f2988ddaffaac47f2c95f378102dad490e911de157/coredis-4.17.0-cp312-cp312-macosx_10_9_x86_64.whl", hash = "sha256:e3638c9a894ac7d0a04fa14515f24d0f717c431266ee0ac612ddb3a142862258", size = 330509 },
+    { url = "https://files.pythonhosted.org/packages/4c/2d/1f97441d377b457831bd9327dbdaa29888effa2edf6318cb4138a425538f/coredis-4.17.0-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:73cb260bf96eacb4e455c300b5e41382bc52d9a2125f3f7e55657662a627e0cb", size = 327735 },
+    { url = "https://files.pythonhosted.org/packages/3a/3f/1dcd57f6df67b7a20b1c27abcf768cf6789be5f33d173739f482d672e9d1/coredis-4.17.0-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:9421423bb109eb62b7595e1d0c84d8c9399bf160826ee478b6b7771bf6ad831e", size = 353755 },
+    { url = "https://files.pythonhosted.org/packages/38/24/de68bdd4b3549a8a05674f0952e646d45afd15453543e0e679dc6899174c/coredis-4.17.0-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:a74abdeda89ff5ea40d0da771d2871148b64b2f1c758f11485397adc1928b08e", size = 357309 },
+    { url = "https://files.pythonhosted.org/packages/ab/66/2bd9f9e1c10b307caf8f4e77527c620a0320291aa83a9e0e98e8df5a326c/coredis-4.17.0-cp312-cp312-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:0ddad826c5bc91f05e5fe36435086cdbe51019b2f4f0faf96d40250823548fee", size = 360856 },
+    { url = "https://files.pythonhosted.org/packages/08/1c/7249845c0f6105290d70d90c9ad48b550f5bcb989766819d38aa0f784aec/coredis-4.17.0-py3-none-any.whl", hash = "sha256:a8254fcc746efd72990d565d87e5399646ad737b7a61d86ef129df846e86b0d3", size = 239667 },
+]
+
 [[package]]
 name = "coverage"
 version = "7.6.9"
@@ -507,6 +547,18 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/d5/50/83c593b07763e1161326b3b8c6686f0f4b0f24d5526546bee538c89837d6/decorator-5.1.1-py3-none-any.whl", hash = "sha256:b8c3f85900b9dc423225913c5aace94729fe1fa9763b38939a95226f02d37186", size = 9073 },
 ]
 
+[[package]]
+name = "deprecated"
+version = "1.2.15"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "wrapt" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/2e/a3/53e7d78a6850ffdd394d7048a31a6f14e44900adedf190f9a165f6b69439/deprecated-1.2.15.tar.gz", hash = "sha256:683e561a90de76239796e6b6feac66b99030d2dd3fcf61ef996330f14bbb9b0d", size = 2977612 }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/1d/8f/c7f227eb42cfeaddce3eb0c96c60cbca37797fa7b34f8e1aeadf6c5c0983/Deprecated-1.2.15-py2.py3-none-any.whl", hash = "sha256:353bc4a8ac4bfc96800ddab349d89c25dec1079f65fd53acdcc1e0b975b21320", size = 9941 },
+]
+
 [[package]]
 name = "dicttoxml"
 version = "1.7.16"
@@ -632,6 +684,24 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/52/b3/7e4df40e585df024fac2f80d1a2d579c854ac37109675db2b0cc22c0bb9e/fastapi-0.115.6-py3-none-any.whl", hash = "sha256:e9240b29e36fa8f4bb7290316988e90c381e5092e0cbe84e7818cc3713bcf305", size = 94843 },
 ]
 
+[[package]]
+name = "fh-llm-client"
+version = "0.0.6"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "coredis" },
+    { name = "fhaviary" },
+    { name = "limits" },
+    { name = "litellm" },
+    { name = "pydantic" },
+    { name = "tiktoken" },
+    { name = "typing-extensions", marker = "python_full_version < '3.12'" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/e8/ba/0395955b6dc64bdfe5724f0b5d8a8e8c916454d1fa0000c0686cd8432d4a/fh_llm_client-0.0.6.tar.gz", hash = "sha256:71ffa711f1bfe2c16b422ee4762a668e443c8351c485924ceb6745db4359dab0", size = 195993 }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/30/de/ee766cd71dec5a12cbb0c7bc8d8a90764bde730376ddee466902f239e756/fh_llm_client-0.0.6-py3-none-any.whl", hash = "sha256:7c20364b465db784c0320132c216cadd933fb56972c7bdff3996bd4a14769926", size = 31801 },
+]
+
 [[package]]
 name = "fhaviary"
 version = "0.11.0"
@@ -1112,14 +1182,14 @@ wheels = [
 
 [[package]]
 name = "ldp"
-version = "0.14.6.dev9+g261607d"
+version = "0.14.4.dev26+gc677c6b.d20241211"
 source = { editable = "." }
 dependencies = [
     { name = "aiofiles" },
     { name = "dm-tree" },
+    { name = "fh-llm-client" },
     { name = "fhaviary" },
     { name = "httpx" },
-    { name = "litellm" },
     { name = "networkx", extra = ["default"] },
     { name = "numpy" },
     { name = "pydantic" },
@@ -1184,10 +1254,62 @@ visualization = [
 [package.dev-dependencies]
 codeflash = [
     { name = "codeflash" },
-    { name = "ldp", extra = ["dev"] },
+    { name = "fastapi" },
+    { name = "fhaviary", extra = ["xml"] },
+    { name = "ipython" },
+    { name = "litellm" },
+    { name = "mypy" },
+    { name = "pre-commit" },
+    { name = "pydantic" },
+    { name = "pydot" },
+    { name = "pylint" },
+    { name = "pylint-pydantic" },
+    { name = "pytest" },
+    { name = "pytest-asyncio" },
+    { name = "pytest-mock" },
+    { name = "pytest-recording" },
+    { name = "pytest-rerunfailures" },
+    { name = "pytest-subtests" },
+    { name = "pytest-sugar" },
+    { name = "pytest-timer", extra = ["colorama"] },
+    { name = "pytest-xdist" },
+    { name = "refurb" },
+    { name = "rich" },
+    { name = "torch" },
+    { name = "tqdm" },
+    { name = "types-aiofiles" },
+    { name = "types-tqdm" },
+    { name = "vcrpy" },
+    { name = "wandb" },
 ]
 dev = [
-    { name = "ldp", extra = ["dev"] },
+    { name = "fastapi" },
+    { name = "fhaviary", extra = ["xml"] },
+    { name = "ipython" },
+    { name = "litellm" },
+    { name = "mypy" },
+    { name = "pre-commit" },
+    { name = "pydantic" },
+    { name = "pydot" },
+    { name = "pylint" },
+    { name = "pylint-pydantic" },
+    { name = "pytest" },
+    { name = "pytest-asyncio" },
+    { name = "pytest-mock" },
+    { name = "pytest-recording" },
+    { name = "pytest-rerunfailures" },
+    { name = "pytest-subtests" },
+    { name = "pytest-sugar" },
+    { name = "pytest-timer", extra = ["colorama"] },
+    { name = "pytest-xdist" },
+    { name = "refurb" },
+    { name = "rich" },
+    { name = "torch" },
+    { name = "tqdm" },
+    { name = "types-aiofiles" },
+    { name = "types-tqdm" },
+    { name = "vcrpy" },
+    { name = "wandb" },
 ]
 
 [package.metadata]
@@ -1195,13 +1317,13 @@ requires-dist = [
     { name = "aiofiles" },
     { name = "dm-tree" },
     { name = "fastapi", marker = "extra == 'server'", specifier = ">=0.109" },
+    { name = "fh-llm-client" },
     { name = "fhaviary", specifier = ">=0.8.2" },
     { name = "fhaviary", extras = ["xml"], marker = "extra == 'dev'" },
     { name = "httpx" },
     { name = "ipython", marker = "extra == 'dev'", specifier = ">=8" },
     { name = "ldp", extras = ["monitor", "nn", "rich", "server", "typing", "visualization"], marker = "extra == 'dev'" },
-    { name = "litellm", specifier = ">=1.52.15" },
-    { name = "litellm", marker = "extra == 'dev'", specifier = "!=1.49.4,!=1.49.5,!=1.49.6" },
+    { name = "litellm", marker = "extra == 'dev'", specifier = ">=1.52.15" },
     { name = "mypy", marker = "extra == 'dev'", specifier = ">=1.8" },
     { name = "networkx", extras = ["default"], specifier = "~=3.4" },
     { name = "numpy", specifier = ">=1.20" },
@@ -1277,6 +1399,20 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/53/54/3bf26fc2cd9f3b5a1ce4f574ddd1238b96d6e4a9c1c0fccd0c10f7e84e59/libcst-1.5.1-cp313-cp313-win_amd64.whl", hash = "sha256:b5a0d3c632aa2b21c5fa145e4e8dbf86f45c9b37a64c0b7221a5a45caf58915a", size = 2032277 },
 ]
 
+[[package]]
+name = "limits"
+version = "3.14.1"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "deprecated" },
+    { name = "packaging" },
+    { name = "typing-extensions" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/18/2f/686912cae32c690ced2cda741965d717b4d376679d7202502000972a5443/limits-3.14.1.tar.gz", hash = "sha256:cad16a9b3cf3924e27da48e78bdab33ef312ecb7194fdb50e509cc8111c8d0bb", size = 70368 }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/b3/5d/150c6f0b3a9e20acd785a6f27e1b2959b136483d39b2ac0a6ae87d3b3f92/limits-3.14.1-py3-none-any.whl", hash = "sha256:051aca02da56e6932599a25cb8e70543959294f5d587d57bcd7e38df234e697b", size = 45651 },
+]
+
 [[package]]
 name = "litellm"
 version = "1.53.3"
@@ -1809,7 +1945,7 @@ wheels = [
 
 [[package]]
 name = "openai"
-version = "1.57.0"
+version = "1.56.2"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
     { name = "anyio" },
@@ -1821,9 +1957,9 @@ dependencies = [
     { name = "tqdm" },
     { name = "typing-extensions" },
 ]
-sdist = { url = "https://files.pythonhosted.org/packages/fa/64/4acd9331b3c0e1069f36692d4c29d2c8deea6649a1e150f45a096f91b339/openai-1.57.0.tar.gz", hash = "sha256:76f91971c4bdbd78380c9970581075e0337b5d497c2fbf7b5255078f4b31abf9", size = 315514 }
+sdist = { url = "https://files.pythonhosted.org/packages/e1/97/302669f5888d1adf8ce0e93f1e5b2337b6c9db47d9877bd344a29db314be/openai-1.56.2.tar.gz", hash = "sha256:17312af69bc7670d4048f98ab5849f8784d98c39ac64fcde19406e3774a0c1e5", size = 315404 }
 wheels = [
-    { url = "https://files.pythonhosted.org/packages/ab/2d/eb8539a2d5809eb78508633a8faa8df7745960e99af0388310c43b2c0be1/openai-1.57.0-py3-none-any.whl", hash = "sha256:972e36960b821797952da3dc4532f486c28e28a2a332d7d0c5407f242e9d9c39", size = 389854 },
+    { url = "https://files.pythonhosted.org/packages/23/36/c60fffa518d82952335e2ef4a6b93b0427c57eb49469879dee1cbe59d551/openai-1.56.2-py3-none-any.whl", hash = "sha256:82d0c48f9504e04c7797e9b799dcf7f49a246d99b6cbfd90f3193ea80815b69e", size = 389854 },
 ]
 
 [[package]]
@@ -2132,16 +2268,16 @@ wheels = [
 
 [[package]]
 name = "pydantic"
-version = "2.10.3"
+version = "2.10.1"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
     { name = "annotated-types" },
     { name = "pydantic-core" },
     { name = "typing-extensions" },
 ]
-sdist = { url = "https://files.pythonhosted.org/packages/45/0f/27908242621b14e649a84e62b133de45f84c255eecb350ab02979844a788/pydantic-2.10.3.tar.gz", hash = "sha256:cb5ac360ce894ceacd69c403187900a02c4b20b693a9dd1d643e1effab9eadf9", size = 786486 }
+sdist = { url = "https://files.pythonhosted.org/packages/c4/bd/7fc610993f616d2398958d0028d15eaf53bde5f80cb2edb7aa4f1feaf3a7/pydantic-2.10.1.tar.gz", hash = "sha256:a4daca2dc0aa429555e0656d6bf94873a7dc5f54ee42b1f5873d666fb3f35560", size = 783717 }
 wheels = [
-    { url = "https://files.pythonhosted.org/packages/62/51/72c18c55cf2f46ff4f91ebcc8f75aa30f7305f3d726be3f4ebffb4ae972b/pydantic-2.10.3-py3-none-any.whl", hash = "sha256:be04d85bbc7b65651c5f8e6b9976ed9c6f41782a55524cef079a34a0bb82144d", size = 456997 },
+    { url = "https://files.pythonhosted.org/packages/e0/fc/fda48d347bd50a788dd2a0f318a52160f911b86fc2d8b4c86f4d7c9bceea/pydantic-2.10.1-py3-none-any.whl", hash = "sha256:a8d20db84de64cf4a7d59e899c2caf0fe9d660c7cfc482528e7020d7dd189a7e", size = 455329 },
 ]
 
 [[package]]
@@ -2274,6 +2410,18 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/1c/00/adca2ef68fde34bf6b84870cd3d0e2d3f8e116263628af99e00c4e393111/pylint_pydantic-0.3.4-py3-none-any.whl", hash = "sha256:f82fdf6b05045102fef2bd8b553a118aadf80155116f374a76eb201c47160a68", size = 16122 },
 ]
 
+[[package]]
+name = "pympler"
+version = "1.1"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "pywin32", marker = "platform_system == 'Windows'" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/dd/37/c384631908029676d8e7213dd956bb686af303a80db7afbc9be36bc49495/pympler-1.1.tar.gz", hash = "sha256:1eaa867cb8992c218430f1708fdaccda53df064144d1c5656b1e6f1ee6000424", size = 179954 }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/79/4f/a6a2e2b202d7fd97eadfe90979845b8706676b41cbd3b42ba75adf329d1f/Pympler-1.1-py3-none-any.whl", hash = "sha256:5b223d6027d0619584116a0cbc28e8d2e378f7a79c1e5e024f9ff3b673c58506", size = 165766 },
+]
+
 [[package]]
 name = "pyparsing"
 version = "3.2.0"
@@ -2447,6 +2595,22 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/11/c3/005fcca25ce078d2cc29fd559379817424e94885510568bc1bc53d7d5846/pytz-2024.2-py2.py3-none-any.whl", hash = "sha256:31c7c1817eb7fae7ca4b8c7ee50c72f93aa2dd863de768e1ef4245d426aa0725", size = 508002 },
 ]
 
+[[package]]
+name = "pywin32"
+version = "308"
+source = { registry = "https://pypi.org/simple" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/eb/e2/02652007469263fe1466e98439831d65d4ca80ea1a2df29abecedf7e47b7/pywin32-308-cp311-cp311-win32.whl", hash = "sha256:5d8c8015b24a7d6855b1550d8e660d8daa09983c80e5daf89a273e5c6fb5095a", size = 5928156 },
+    { url = "https://files.pythonhosted.org/packages/48/ef/f4fb45e2196bc7ffe09cad0542d9aff66b0e33f6c0954b43e49c33cad7bd/pywin32-308-cp311-cp311-win_amd64.whl", hash = "sha256:575621b90f0dc2695fec346b2d6302faebd4f0f45c05ea29404cefe35d89442b", size = 6559559 },
+    { url = "https://files.pythonhosted.org/packages/79/ef/68bb6aa865c5c9b11a35771329e95917b5559845bd75b65549407f9fc6b4/pywin32-308-cp311-cp311-win_arm64.whl", hash = "sha256:100a5442b7332070983c4cd03f2e906a5648a5104b8a7f50175f7906efd16bb6", size = 7972495 },
+    { url = "https://files.pythonhosted.org/packages/00/7c/d00d6bdd96de4344e06c4afbf218bc86b54436a94c01c71a8701f613aa56/pywin32-308-cp312-cp312-win32.whl", hash = "sha256:587f3e19696f4bf96fde9d8a57cec74a57021ad5f204c9e627e15c33ff568897", size = 5939729 },
+    { url = "https://files.pythonhosted.org/packages/21/27/0c8811fbc3ca188f93b5354e7c286eb91f80a53afa4e11007ef661afa746/pywin32-308-cp312-cp312-win_amd64.whl", hash = "sha256:00b3e11ef09ede56c6a43c71f2d31857cf7c54b0ab6e78ac659497abd2834f47", size = 6543015 },
+    { url = "https://files.pythonhosted.org/packages/9d/0f/d40f8373608caed2255781a3ad9a51d03a594a1248cd632d6a298daca693/pywin32-308-cp312-cp312-win_arm64.whl", hash = "sha256:9b4de86c8d909aed15b7011182c8cab38c8850de36e6afb1f0db22b8959e3091", size = 7976033 },
+    { url = "https://files.pythonhosted.org/packages/a9/a4/aa562d8935e3df5e49c161b427a3a2efad2ed4e9cf81c3de636f1fdddfd0/pywin32-308-cp313-cp313-win32.whl", hash = "sha256:1c44539a37a5b7b21d02ab34e6a4d314e0788f1690d65b48e9b0b89f31abbbed", size = 5938579 },
+    { url = "https://files.pythonhosted.org/packages/c7/50/b0efb8bb66210da67a53ab95fd7a98826a97ee21f1d22949863e6d588b22/pywin32-308-cp313-cp313-win_amd64.whl", hash = "sha256:fd380990e792eaf6827fcb7e187b2b4b1cede0585e3d0c9e84201ec27b9905e4", size = 6542056 },
+    { url = "https://files.pythonhosted.org/packages/26/df/2b63e3e4f2df0224f8aaf6d131f54fe4e8c96400eb9df563e2aae2e1a1f9/pywin32-308-cp313-cp313-win_arm64.whl", hash = "sha256:ef313c46d4c18dfb82a2431e3051ac8f112ccee1a34f29c263c583c568db63cd", size = 7974986 },
+]
+
 [[package]]
 name = "pyyaml"
 version = "6.0.2"
@@ -3058,7 +3222,7 @@ name = "triton"
 version = "3.1.0"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
-    { name = "filelock" },
+    { name = "filelock", marker = "python_full_version < '3.13'" },
 ]
 wheels = [
     { url = "https://files.pythonhosted.org/packages/86/17/d9a5cf4fcf46291856d1e90762e36cbabd2a56c7265da0d1d9508c8e3943/triton-3.1.0-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:0f34f6e7885d1bf0eaaf7ba875a5f0ce6f3c13ba98f9503651c1e6dc6757ed5c", size = 209506424 },