diff --git a/fastdeploy/engine/args_utils.py b/fastdeploy/engine/args_utils.py
index c6811e351..306423aaf 100644
--- a/fastdeploy/engine/args_utils.py
+++ b/fastdeploy/engine/args_utils.py
@@ -398,6 +398,8 @@ class EngineArgs:
             envs.ENABLE_V1_KVCACHE_SCHEDULER = 0
         if not current_platform.is_cuda():
             envs.ENABLE_V1_KVCACHE_SCHEDULER = 0
+        if self.guided_decoding_backend != "off":
+            envs.ENABLE_V1_KVCACHE_SCHEDULER = 0
 
     @staticmethod
     def add_cli_args(parser: FlexibleArgumentParser) -> FlexibleArgumentParser:
diff --git a/fastdeploy/worker/worker_process.py b/fastdeploy/worker/worker_process.py
index 8ba805023..8a0ff6f09 100644
--- a/fastdeploy/worker/worker_process.py
+++ b/fastdeploy/worker/worker_process.py
@@ -755,6 +755,9 @@ def initialize_fd_config(args, ranks: int = 1, local_rank: int = 0) -> FDConfig:
     if not current_platform.is_cuda():
         logger.info("Set ENABLE_V1_KVCACHE_SCHEDULER to 0 due to not supported.")
         envs.ENABLE_V1_KVCACHE_SCHEDULER = 0
+    if parallel_config.guided_decoding_backend != "off":
+        logger.info("Set ENABLE_V1_KVCACHE_SCHEDULER to 0 due to not supported guided_decoding.")
+        envs.ENABLE_V1_KVCACHE_SCHEDULER = 0
 
     fd_config = FDConfig(
         model_config=model_config,