diff --git a/agentneo/utils/trace_utils.py b/agentneo/utils/trace_utils.py
index 0cd9a669..e414a763 100644
--- a/agentneo/utils/trace_utils.py
+++ b/agentneo/utils/trace_utils.py
@@ -1,5 +1,6 @@
 import json
 from importlib import resources
+from dataclasses import asdict
 
 
 def convert_usage_to_dict(usage):
diff --git a/tests/test_utils/test_trace_utils.py b/tests/test_utils/test_trace_utils.py
new file mode 100644
index 00000000..f50a2ab7
--- /dev/null
+++ b/tests/test_utils/test_trace_utils.py
@@ -0,0 +1,72 @@
+import pytest
+import json
+from dataclasses import dataclass
+from agentneo.utils import trace_utils
+from unittest.mock import mock_open, patch
+
+# Define a mock data class for testing log_event
+@dataclass
+class MockEvent:
+    event_type: str
+    description: str
+
+# Test convert_usage_to_dict function
+def test_convert_usage_to_dict():
+    # Test with a dictionary input
+    usage_dict = {"prompt_tokens": 5, "completion_tokens": 10, "reasoning_tokens": 2}
+    expected_dict = {"input": 5, "completion": 10, "reasoning": 2}
+    assert trace_utils.convert_usage_to_dict(usage_dict) == expected_dict
+
+    # Test with an object with attributes
+    class UsageObject:
+        prompt_tokens = 8
+        completion_tokens = 4
+        reasoning_tokens = 1
+
+    usage_object = UsageObject()
+    expected_object_dict = {"input": 8, "completion": 4, "reasoning": 1}
+    assert trace_utils.convert_usage_to_dict(usage_object) == expected_object_dict
+
+    # Test with an unexpected type
+    assert trace_utils.convert_usage_to_dict("invalid_type") == {"input": 0, "completion": 0, "reasoning": 0}
+
+# Test calculate_cost function
+def test_calculate_cost():
+    token_usage = {"prompt_tokens": 10, "completion_tokens": 5, "reasoning_tokens": 2}
+    input_cost_per_token = 0.01
+    output_cost_per_token = 0.02
+    reasoning_cost_per_token = 0.03
+
+    expected_cost = {
+        "input": 0.1,            # 10 * 0.01
+        "completion": 0.1,       # 5 * 0.02
+        "reasoning": 0.06,       # 2 * 0.03
+        "total": 0.26            # sum of all costs
+    }
+    assert trace_utils.calculate_cost(token_usage, input_cost_per_token, output_cost_per_token, reasoning_cost_per_token) == expected_cost
+
+# Test load_model_costs function
+def test_load_model_costs():
+    # Simulate file found case
+    model_costs_content = {"modelA": {"input_cost": 0.01, "output_cost": 0.02}}
+    with patch("builtins.open", mock_open(read_data=json.dumps(model_costs_content))):
+        assert trace_utils.load_model_costs() == model_costs_content
+
+    # Simulate file not found case, using a fallback file within resources
+    with patch("builtins.open", side_effect=FileNotFoundError):
+        with patch("importlib.resources.open_text", mock_open(read_data=json.dumps(model_costs_content))):
+            assert trace_utils.load_model_costs() == model_costs_content
+
+# Test log_event function
+def test_log_event(tmp_path):
+    # Use a temporary file path for logging
+    log_file_path = tmp_path / "log.txt"
+    event = MockEvent(event_type="INFO", description="Test event")
+    
+    trace_utils.log_event(event, log_file_path)
+    
+    # Read the log file and verify its contents
+    with open(log_file_path, "r") as f:
+        log_entry = json.loads(f.readline())
+        assert log_entry["event_type"] == "INFO"
+        assert log_entry["description"] == "Test event"