diff --git a/src/constants.h b/src/constants.h
index 8415f8ee9..119d1e9d2 100644
--- a/src/constants.h
+++ b/src/constants.h
@@ -1,4 +1,4 @@
-// Copyright 2018-2024, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+// Copyright 2018-2025, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 //
 // Redistribution and use in source and binary forms, with or without
 // modification, are permitted provided that the following conditions
@@ -92,7 +92,6 @@ constexpr uint64_t NANOS_PER_SECOND = 1000000000;
 constexpr uint64_t NANOS_PER_MILLIS = 1000000;
 constexpr int MAX_GRPC_MESSAGE_SIZE = INT32_MAX;
 constexpr uint64_t SEQUENCE_IDLE_DEFAULT_MICROSECONDS = 1000 * 1000;
-constexpr size_t STRING_CORRELATION_ID_MAX_LENGTH_BYTES = 128;
 constexpr size_t CUDA_IPC_STRUCT_SIZE = 64;
 
 #ifdef TRITON_ENABLE_METRICS
diff --git a/src/infer_request.cc b/src/infer_request.cc
index 0d0c80a0d..83b3bb872 100644
--- a/src/infer_request.cc
+++ b/src/infer_request.cc
@@ -1,4 +1,4 @@
-// Copyright 2020-2024, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+// Copyright 2020-2025, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 //
 // Redistribution and use in source and binary forms, with or without
 // modification, are permitted provided that the following conditions
@@ -1228,11 +1228,16 @@ InferenceRequest::Normalize()
       }
     }
   }
+
+  if (model_config.has_sequence_batching()) {
+    RETURN_IF_ERROR(ValidateCorrelationId());
+  }
+
   return Status::Success;
 }
 
 Status
-InferenceRequest::ValidateRequestInputs()
+InferenceRequest::ValidateRequestInputs() const
 {
   const inference::ModelConfig& model_config = model_raw_->Config();
   if ((original_inputs_.size() > (size_t)model_config.input_size()) ||
@@ -1404,6 +1409,59 @@ InferenceRequest::ValidateBytesInputs(
   return Status::Success;
 }
 
+Status
+InferenceRequest::ValidateCorrelationId() const
+{
+  const inference::ModelConfig& model_config = model_raw_->Config();
+  const std::string& model_name = ModelName();
+  std::string correlation_id_tensor_name;
+  inference::DataType correlation_id_datatype;
+
+  RETURN_IF_ERROR(GetTypedSequenceControlProperties(
+      model_config.sequence_batching(), model_config.name(),
+      inference::ModelSequenceBatching::Control::CONTROL_SEQUENCE_CORRID,
+      false /* required */, &correlation_id_tensor_name,
+      &correlation_id_datatype));
+
+  // Make sure request correlation ID type matches model configuration.
+  if (!correlation_id_tensor_name.empty()) {
+    const auto& correlation_id = CorrelationId();
+    bool dtypes_match = true;
+    std::string request_corrid_datatype;
+    if ((correlation_id.Type() ==
+         InferenceRequest::SequenceId::DataType::STRING) &&
+        (correlation_id_datatype != inference::DataType::TYPE_STRING)) {
+      dtypes_match = false;
+      request_corrid_datatype = triton::common::DataTypeToProtocolString(
+          inference::DataType::TYPE_STRING);
+    } else if (
+        (correlation_id.Type() ==
+         InferenceRequest::SequenceId::DataType::UINT64) &&
+        ((correlation_id_datatype != inference::DataType::TYPE_UINT64) &&
+         (correlation_id_datatype != inference::DataType::TYPE_INT64) &&
+         (correlation_id_datatype != inference::DataType::TYPE_UINT32) &&
+         (correlation_id_datatype != inference::DataType::TYPE_INT32))) {
+      dtypes_match = false;
+      request_corrid_datatype = triton::common::DataTypeToProtocolString(
+          inference::DataType::TYPE_UINT64);
+    }
+
+    if (!dtypes_match) {
+      return Status(
+          Status::Code::INVALID_ARG,
+          LogRequest() + "sequence batching control '" +
+              correlation_id_tensor_name + "' data-type is '" +
+              request_corrid_datatype + "', but model '" + model_name +
+              "' expects '" +
+              std::string(triton::common::DataTypeToProtocolString(
+                  correlation_id_datatype)) +
+              "'");
+    }
+  }
+
+  return Status::Success;
+}
+
 #ifdef TRITON_ENABLE_STATS
 
 void
diff --git a/src/infer_request.h b/src/infer_request.h
index 38c89ed63..e9bfa49bc 100644
--- a/src/infer_request.h
+++ b/src/infer_request.h
@@ -1,4 +1,4 @@
-// Copyright 2020-2024, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+// Copyright 2020-2025, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 //
 // Redistribution and use in source and binary forms, with or without
 // modification, are permitted provided that the following conditions
@@ -771,13 +771,15 @@ class InferenceRequest {
   Status Normalize();
 
   // Helper for validating Inputs
-  Status ValidateRequestInputs();
+  Status ValidateRequestInputs() const;
 
   Status ValidateBytesInputs(
       const std::string& input_id, const Input& input,
       const std::string& model_name,
       TRITONSERVER_MemoryType* buffer_memory_type) const;
 
+  Status ValidateCorrelationId() const;
+
   // Helpers for pending request metrics
   void IncrementPendingRequestCount();
   void DecrementPendingRequestCount();
diff --git a/src/sequence_batch_scheduler/sequence_batch_scheduler.cc b/src/sequence_batch_scheduler/sequence_batch_scheduler.cc
index 74314e7ab..45e9c037c 100644
--- a/src/sequence_batch_scheduler/sequence_batch_scheduler.cc
+++ b/src/sequence_batch_scheduler/sequence_batch_scheduler.cc
@@ -1,4 +1,4 @@
-// Copyright 2018-2024, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+// Copyright 2018-2025, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 //
 // Redistribution and use in source and binary forms, with or without
 // modification, are permitted provided that the following conditions
@@ -1343,9 +1343,9 @@ SequenceBatch::SetControlTensors(
     auto& seq_corr_id = seq_slot_corrid_override_;
     size_t size_p = triton::common::GetDataTypeByteSize(seq_corr_id->DType());
     if (seq_corr_id->DType() == inference::DataType::TYPE_STRING) {
-      // 4 bytes for length of string plus pre-defined max string correlation id
-      // length in bytes
-      size_p = 4 + triton::core::STRING_CORRELATION_ID_MAX_LENGTH_BYTES;
+      // 4 bytes for length of string plus string correlation id length in
+      // bytes.
+      size_p = 4 + corrid.StringValue().length();
     }
 
     TRITONSERVER_MemoryType memory_type;