diff --git a/fastdeploy/envs.py b/fastdeploy/envs.py
index 23030d6a8..7c6eb5941 100644
--- a/fastdeploy/envs.py
+++ b/fastdeploy/envs.py
@@ -153,8 +153,12 @@ environment_variables: dict[str, Callable[[], Any]] = {
     "FD_MODEL_USE_SAFETENSORS": lambda: int(os.getenv("FD_MODEL_USE_SAFETENSORS", "1")) == 1,
     "FD_MODEL_USE_OFFLINE_QUANT": lambda: int(os.getenv("FD_MODEL_USE_OFFLINE_QUANT", "1")) == 1,
     "FD_MOE_QUANT_TYPE": lambda: os.getenv("FD_MOE_QUANT_TYPE", "w4a8"),
+    # The AK of bos storing the features while multi_modal infer
     "ENCODE_FEATURE_BOS_AK": lambda: os.getenv("ENCODE_FEATURE_BOS_AK"),
+    # The SK of bos storing the features while multi_modal infer
     "ENCODE_FEATURE_BOS_SK": lambda: os.getenv("ENCODE_FEATURE_BOS_SK"),
+    # The ENDPOINT of bos storing the features while multi_modal infer
+    "ENCODE_FEATURE_ENDPOINT": lambda: os.getenv("ENCODE_FEATURE_ENDPOINT"),
     # Enable offline perf test mode for PD disaggregation
     "FD_OFFLINE_PERF_TEST_FOR_PD": lambda: int(os.getenv("FD_OFFLINE_PERF_TEST_FOR_PD", "0")),
     "FD_ENABLE_E2W_TENSOR_CONVERT": lambda: int(os.getenv("FD_ENABLE_E2W_TENSOR_CONVERT", "0")),
diff --git a/fastdeploy/utils.py b/fastdeploy/utils.py
index 17838db6e..492f23b0e 100644
--- a/fastdeploy/utils.py
+++ b/fastdeploy/utils.py
@@ -972,7 +972,8 @@ def init_bos_client():
     from baidubce.services.bos.bos_client import BosClient
 
     cfg = BceClientConfiguration(
-        credentials=BceCredentials(envs.ENCODE_FEATURE_BOS_AK, envs.ENCODE_FEATURE_BOS_SK), endpoint="bj.bcebos.com"
+        credentials=BceCredentials(envs.ENCODE_FEATURE_BOS_AK, envs.ENCODE_FEATURE_BOS_SK),
+        endpoint=envs.ENCODE_FEATURE_ENDPOINT,
     )
     return BosClient(cfg)