diff --git a/tensorrt_llm/_torch/pyexecutor/kv_cache_connector.py b/tensorrt_llm/_torch/pyexecutor/kv_cache_connector.py
index 59f7256c98..380486e935 100644
--- a/tensorrt_llm/_torch/pyexecutor/kv_cache_connector.py
+++ b/tensorrt_llm/_torch/pyexecutor/kv_cache_connector.py
@@ -49,6 +49,7 @@ from tensorrt_llm.bindings.internal.batch_manager import \
 from tensorrt_llm.bindings.internal.batch_manager import LlmRequest
 from tensorrt_llm.llmapi.llm_args import TorchLlmArgs
 
+from .llm_request import get_draft_token_length
 from .scheduler import ScheduledRequests
 
 if TYPE_CHECKING:
@@ -310,7 +311,8 @@ class KvCacheConnectorSchedulerOutputRequest:
                                        req.context_chunk_size)
         else:
             computed_position = len(tokens) - 1
-            num_scheduled_tokens = 1  # Specdec with draft tokens is not supported yet.
+            num_scheduled_tokens = 1 + get_draft_token_length(
+                req)  # Specdec with draft tokens is not supported yet.
 
         return RequestData(req.request_id, new_tokens, new_block_ids,
                            computed_position, num_scheduled_tokens)
diff --git a/tensorrt_llm/_torch/pyexecutor/py_executor_creator.py b/tensorrt_llm/_torch/pyexecutor/py_executor_creator.py
index 46960dabe7..50feb71943 100644
--- a/tensorrt_llm/_torch/pyexecutor/py_executor_creator.py
+++ b/tensorrt_llm/_torch/pyexecutor/py_executor_creator.py
@@ -554,9 +554,6 @@ def create_py_executor(
             raise NotImplementedError(
                 "KV connector is only supported with guaranteed no evict scheduler policy."
             )
-        elif spec_config is not None:
-            raise NotImplementedError(
-                "KV connector is not supported with speculative decoding.")
         try:
             module = importlib.import_module(
                 kv_connector_config.connector_module)