diff --git a/GPU/GPUTracking/Base/GPUParam.cxx b/GPU/GPUTracking/Base/GPUParam.cxx
index 17fcf6d2c9a72..1b67cbd26d45c 100644
--- a/GPU/GPUTracking/Base/GPUParam.cxx
+++ b/GPU/GPUTracking/Base/GPUParam.cxx
@@ -120,6 +120,7 @@ void GPUParam::SetDefaults(float solenoidBz)
   par.toyMCEventsFlag = false;
   par.continuousTracking = false;
   continuousMaxTimeBin = 0;
+  tpcCutTimeBin = 0;
   par.debugLevel = 0;
   par.earlyTpcTransform = false;
 }
@@ -132,6 +133,7 @@ void GPUParam::UpdateSettings(const GPUSettingsGRP* g, const GPUSettingsProcessi
     par.toyMCEventsFlag = g->homemadeEvents;
     par.continuousTracking = g->grpContinuousMaxTimeBin != 0;
     continuousMaxTimeBin = g->grpContinuousMaxTimeBin == -1 ? GPUSettings::TPC_MAX_TF_TIME_BIN : g->grpContinuousMaxTimeBin;
+    tpcCutTimeBin = g->tpcCutTimeBin;
   }
   par.earlyTpcTransform = rec.tpc.forceEarlyTransform == -1 ? (!par.continuousTracking) : rec.tpc.forceEarlyTransform;
   qptB5Scaler = CAMath::Abs(bzkG) > 0.1f ? CAMath::Abs(bzkG) / 5.006680f : 1.f; // Repeat here, since passing in g is optional
diff --git a/GPU/GPUTracking/Base/GPUParam.h b/GPU/GPUTracking/Base/GPUParam.h
index fd380c0a39593..ce9ac30b7c35b 100644
--- a/GPU/GPUTracking/Base/GPUParam.h
+++ b/GPU/GPUTracking/Base/GPUParam.h
@@ -59,6 +59,7 @@ struct GPUParam_t {
 
   int8_t dodEdxDownscaled;
   int32_t continuousMaxTimeBin;
+  int32_t tpcCutTimeBin;
 
   GPUTPCGeometry tpcGeometry;                       // TPC Geometry
   GPUTPCGMPolynomialField polynomialField;          // Polynomial approx. of magnetic field for TPC GM
diff --git a/GPU/GPUTracking/DataTypes/GPUNewCalibValues.cxx b/GPU/GPUTracking/DataTypes/GPUNewCalibValues.cxx
index f443809d15ef5..e86955d6da500 100644
--- a/GPU/GPUTracking/DataTypes/GPUNewCalibValues.cxx
+++ b/GPU/GPUTracking/DataTypes/GPUNewCalibValues.cxx
@@ -24,4 +24,7 @@ void GPUNewCalibValues::updateFrom(const GPUNewCalibValues* from)
   if (from->newContinuousMaxTimeBin) {
     continuousMaxTimeBin = from->continuousMaxTimeBin;
   }
+  if (from->newTPCTimeBinCut) {
+    tpcTimeBinCut = from->tpcTimeBinCut;
+  }
 }
diff --git a/GPU/GPUTracking/DataTypes/GPUNewCalibValues.h b/GPU/GPUTracking/DataTypes/GPUNewCalibValues.h
index 802306e996553..5d5a31785928c 100644
--- a/GPU/GPUTracking/DataTypes/GPUNewCalibValues.h
+++ b/GPU/GPUTracking/DataTypes/GPUNewCalibValues.h
@@ -25,8 +25,10 @@ namespace gpu
 struct GPUNewCalibValues {
   bool newSolenoidField = false;
   bool newContinuousMaxTimeBin = false;
+  bool newTPCTimeBinCut = false;
   float solenoidField = 0.f;
   uint32_t continuousMaxTimeBin = 0;
+  int32_t tpcTimeBinCut = 0;
 
   void updateFrom(const GPUNewCalibValues* from);
 };
diff --git a/GPU/GPUTracking/DataTypes/GPUSettings.h b/GPU/GPUTracking/DataTypes/GPUSettings.h
index 69bfb15e3f4b0..b967a7ce42620 100644
--- a/GPU/GPUTracking/DataTypes/GPUSettings.h
+++ b/GPU/GPUTracking/DataTypes/GPUSettings.h
@@ -60,6 +60,7 @@ struct GPUSettingsGRP {
   int32_t grpContinuousMaxTimeBin = -2;   // 0 for triggered events, -1 for automatic setting, -2 invalid default
   int32_t needsClusterer = 0;             // Set to true if the data requires the clusterizer
   int32_t doCompClusterDecode = 0;        // Set to true if the data contains compressed TPC clusters
+  int32_t tpcCutTimeBin = 0;              // Cut TPC clusters and digits >= this cut
 };
 
 // Parameters of the current time frame
diff --git a/GPU/GPUTracking/Global/GPUChainTracking.cxx b/GPU/GPUTracking/Global/GPUChainTracking.cxx
index 8c2599604387b..abb44ec31ae22 100644
--- a/GPU/GPUTracking/Global/GPUChainTracking.cxx
+++ b/GPU/GPUTracking/Global/GPUChainTracking.cxx
@@ -641,7 +641,9 @@ int32_t GPUChainTracking::DoQueuedUpdates(int32_t stream, bool updateSlave)
       if (mNewCalibValues->newContinuousMaxTimeBin) {
         grp->grpContinuousMaxTimeBin = mNewCalibValues->continuousMaxTimeBin;
       }
-    }
+      if (mNewCalibValues->newTPCTimeBinCut) {
+        grp->tpcCutTimeBin = mNewCalibValues->tpcTimeBinCut;
+      }
   }
   if (GetProcessingSettings().tpcDownscaledEdx != 0) {
     p = &GetProcessingSettings();
diff --git a/GPU/GPUTracking/Global/GPUChainTrackingCompression.cxx b/GPU/GPUTracking/Global/GPUChainTrackingCompression.cxx
index 98109447de034..46ef8bb23fcce 100644
--- a/GPU/GPUTracking/Global/GPUChainTrackingCompression.cxx
+++ b/GPU/GPUTracking/Global/GPUChainTrackingCompression.cxx
@@ -225,8 +225,8 @@ int32_t GPUChainTracking::RunTPCDecompression()
       GPUError("Error decompressing clusters");
       return 1;
     }
-    if (GetProcessingSettings().tpcApplyCFCutsAtDecoding) {
-      RunTPCClusterFilter(mClusterNativeAccess.get(), allocatorFinal, true);
+    if (GetProcessingSettings().tpcApplyCFCutsAtDecoding || param().tpcCutTimeBin > 0) {
+      RunTPCClusterFilter(mClusterNativeAccess.get(), allocatorFinal, GetProcessingSettings().tpcApplyCFCutsAtDecoding);
     }
     decompressTimer.Stop();
     mIOPtrs.clustersNative = mClusterNativeAccess.get();
diff --git a/GPU/GPUTracking/Global/GPUChainTrackingDebugAndProfiling.cxx b/GPU/GPUTracking/Global/GPUChainTrackingDebugAndProfiling.cxx
index f8a64e9d4faaa..5cd86277245d1 100644
--- a/GPU/GPUTracking/Global/GPUChainTrackingDebugAndProfiling.cxx
+++ b/GPU/GPUTracking/Global/GPUChainTrackingDebugAndProfiling.cxx
@@ -315,6 +315,9 @@ void GPUChainTracking::RunTPCClusterFilter(o2::tpc::ClusterNativeAccess* cluster
             keep = keep && cl.qTot > param().rec.tpc.cfQTotCutoff && cl.qMax > param().rec.tpc.cfQMaxCutoff;
             keep = keep && (!(cl.getFlags() & o2::tpc::ClusterNative::flagSingle) || ((cl.sigmaPadPacked || cl.qMax > param().rec.tpc.cfQMaxCutoffSinglePad) && (cl.sigmaTimePacked || cl.qMax > param().rec.tpc.cfQMaxCutoffSingleTime)));
           }
+          if (param().tpcCutTimeBin > 0) {
+            keep = keep & cl.getTime() < param().tpcCutTimeBin;
+          }
           keep = keep && (!GetProcessingSettings().tpcApplyDebugClusterFilter || clusterFilter.filter(iSector, iRow, cl));
           if (iPhase && keep) {
             outputBuffer[countTotal] = cl;
diff --git a/GPU/Workflow/include/GPUWorkflow/GPUWorkflowSpec.h b/GPU/Workflow/include/GPUWorkflow/GPUWorkflowSpec.h
index b218a21306a34..eda3b28c6cff6 100644
--- a/GPU/Workflow/include/GPUWorkflow/GPUWorkflowSpec.h
+++ b/GPU/Workflow/include/GPUWorkflow/GPUWorkflowSpec.h
@@ -233,6 +233,7 @@ class GPURecoWorkflowSpec : public o2::framework::Task
   bool mITSGeometryCreated = false;
   bool mTRDGeometryCreated = false;
   bool mPropagatorInstanceCreated = false;
+  int32_t mTPCCutAtTimeBin = -1;
 };
 
 } // end namespace gpu
diff --git a/GPU/Workflow/src/GPUWorkflowSpec.cxx b/GPU/Workflow/src/GPUWorkflowSpec.cxx
index 37ae734845667..06942eab476c6 100644
--- a/GPU/Workflow/src/GPUWorkflowSpec.cxx
+++ b/GPU/Workflow/src/GPUWorkflowSpec.cxx
@@ -1013,9 +1013,8 @@ void GPURecoWorkflowSpec::doCalibUpdates(o2::framework::ProcessingContext& pc, c
       LOG(info) << "Updating solenoid field " << newCalibValues.solenoidField;
     }
     if (mAutoContinuousMaxTimeBin) {
-      mConfig->configGRP.grpContinuousMaxTimeBin = GPUO2InterfaceUtils::getTpcMaxTimeBinFromNHbf(mTFSettings->nHBFPerTF);
       newCalibValues.newContinuousMaxTimeBin = true;
-      newCalibValues.continuousMaxTimeBin = mConfig->configGRP.grpContinuousMaxTimeBin;
+      newCalibValues.continuousMaxTimeBin = mConfig->configGRP.grpContinuousMaxTimeBin = GPUO2InterfaceUtils::getTpcMaxTimeBinFromNHbf(mTFSettings->nHBFPerTF);
       LOG(info) << "Updating max time bin " << newCalibValues.continuousMaxTimeBin << " (" << mTFSettings->nHBFPerTF << " orbits)";
     }
 
@@ -1050,6 +1049,11 @@ void GPURecoWorkflowSpec::doCalibUpdates(o2::framework::ProcessingContext& pc, c
   if (mSpecConfig.runITSTracking) {
     needCalibUpdate = fetchCalibsCCDBITS(pc) || needCalibUpdate;
   }
+  if (mTPCCutAtTimeBin != mConfig->configGRP.tpcCutTimeBin) {
+    newCalibValues.newTPCTimeBinCut = true;
+    newCalibValues.tpcTimeBinCut = mConfig->configGRP.tpcCutTimeBin = mTPCCutAtTimeBin;
+    needCalibUpdate = true;
+  }
   if (needCalibUpdate) {
     LOG(info) << "Updating GPUReconstruction calibration objects";
     mGPUReco->UpdateCalibration(newCalibObjects, newCalibValues);
@@ -1098,6 +1102,7 @@ Inputs GPURecoWorkflowSpec::inputs()
   if (mSpecConfig.outputTracks || mSpecConfig.caClusterer) {
     // calibration objects for TPC clusterization
     inputs.emplace_back("tpcgain", gDataOriginTPC, "PADGAINFULL", 0, Lifetime::Condition, ccdbParamSpec(o2::tpc::CDBTypeMap.at(o2::tpc::CDBType::CalPadGainFull)));
+    inputs.emplace_back("tpcaltrosync", gDataOriginTPC, "ALTROSYNCSIGNAL", 0, Lifetime::Condition, ccdbParamSpec(o2::tpc::CDBTypeMap.at(o2::tpc::CDBType::AltroSyncSignal)));
   }
   if (mSpecConfig.outputTracks) {
     // calibration objects for TPC tracking
diff --git a/GPU/Workflow/src/GPUWorkflowTPC.cxx b/GPU/Workflow/src/GPUWorkflowTPC.cxx
index 97bf3aed26368..ae91716825271 100644
--- a/GPU/Workflow/src/GPUWorkflowTPC.cxx
+++ b/GPU/Workflow/src/GPUWorkflowTPC.cxx
@@ -308,6 +308,10 @@ bool GPURecoWorkflowSpec::fetchCalibsCCDBTPC<GPUCalibObjectsConst>(ProcessingCon
       pc.inputs().get<o2::tpc::CalDet<float>*>("tpcgain");
     }
 
+    if (mSpecConfig.outputTracks || mSpecConfig.caClusterer) {
+      mTPCCutAtTimeBin = pc.inputs().get<o2::tpc::CalDet<o2::tpc::AltroSyncSignal>*>("tpcaltrosync")->getTB2Cut(pc.services().get<o2::framework::TimingInfo>().tfCounter);
+    }
+
     // these calibrations are only defined for the tracking
     if (mSpecConfig.outputTracks) {
       // update the calibration objects in case they changed in the CCDB