diff --git a/python/sglang/srt/managers/detokenizer_manager.py b/python/sglang/srt/managers/detokenizer_manager.py
index 6e90d19cd..4ae31ecc8 100644
--- a/python/sglang/srt/managers/detokenizer_manager.py
+++ b/python/sglang/srt/managers/detokenizer_manager.py
@@ -75,8 +75,8 @@ class DetokenizerManager:
 
         self.decode_status = LimitedCapacityDict()
 
-    def trim_eos(self, output: Union[str, List[int]], finished_reason, no_eos_trim):
-        if no_eos_trim:
+    def trim_eos(self, output: Union[str, List[int]], finished_reason, no_stop_trim):
+        if no_stop_trim:
             return output
 
         # Trim stop str. TODO(lmzheng): handle the case where multiple stop strs are hit
@@ -141,7 +141,7 @@ class DetokenizerManager:
                     self.trim_eos(
                         s.decode_ids[s.surr_offset :],
                         recv_obj.finished_reason[i],
-                        recv_obj.no_eos_trim[i],
+                        recv_obj.no_stop_trim[i],
                     )
                 )
                 surr_ids.append(s.decode_ids[s.surr_offset : s.read_offset])
@@ -177,7 +177,7 @@ class DetokenizerManager:
                     self.trim_eos(
                         s.decoded_text + new_text,
                         recv_obj.finished_reason[i],
-                        recv_obj.no_eos_trim[i],
+                        recv_obj.no_stop_trim[i],
                     )
                 )
 
diff --git a/python/sglang/srt/managers/io_struct.py b/python/sglang/srt/managers/io_struct.py
index 9cc847706..80816c28e 100644
--- a/python/sglang/srt/managers/io_struct.py
+++ b/python/sglang/srt/managers/io_struct.py
@@ -295,7 +295,7 @@ class BatchTokenIDOut:
     spaces_between_special_tokens: List[bool]
     meta_info: List[Dict]
     finished_reason: List[BaseFinishReason]
-    no_eos_trim: List[bool]
+    no_stop_trim: List[bool]
 
 
 @dataclass
diff --git a/python/sglang/srt/managers/scheduler.py b/python/sglang/srt/managers/scheduler.py
index e59679ffa..42c2a2841 100644
--- a/python/sglang/srt/managers/scheduler.py
+++ b/python/sglang/srt/managers/scheduler.py
@@ -885,7 +885,7 @@ class Scheduler:
             output_read_offsets = []
             output_skip_special_tokens = []
             output_spaces_between_special_tokens = []
-            output_no_eos_trim = []
+            output_no_stop_trim = []
         else:  # embedding or reward model
             output_embeddings = []
         unfinished_indices = []
@@ -917,7 +917,7 @@ class Scheduler:
                     output_spaces_between_special_tokens.append(
                         req.sampling_params.spaces_between_special_tokens
                     )
-                    output_no_eos_trim.append(req.sampling_params.no_eos_trim)
+                    output_no_stop_trim.append(req.sampling_params.no_stop_trim)
 
                     meta_info = {
                         "prompt_tokens": len(req.origin_input_ids),
@@ -968,7 +968,7 @@ class Scheduler:
                         output_spaces_between_special_tokens,
                         output_meta_info,
                         output_finished_reason,
-                        output_no_eos_trim,
+                        output_no_stop_trim,
                     )
                 )
             else:  # embedding or reward model
diff --git a/python/sglang/srt/openai_api/adapter.py b/python/sglang/srt/openai_api/adapter.py
index a3638d601..5c8990c69 100644
--- a/python/sglang/srt/openai_api/adapter.py
+++ b/python/sglang/srt/openai_api/adapter.py
@@ -494,7 +494,7 @@ def v1_generate_request(
             request.logprobs if request.logprobs is not None else 0
         )
         sampling_params = []
-        if isinstance(request.no_eos_trim, list):
+        if isinstance(request.no_stop_trim, list):
             num_reqs = len(request.prompt)
         else:
             num_reqs = 1
@@ -514,10 +514,10 @@ def v1_generate_request(
                     "json_schema": request.json_schema,
                     "n": request.n,
                     "ignore_eos": request.ignore_eos,
-                    "no_eos_trim": (
-                        request.no_eos_trim
-                        if not isinstance(request.no_eos_trim, list)
-                        else request.no_eos_trim[i]
+                    "no_stop_trim": (
+                        request.no_stop_trim
+                        if not isinstance(request.no_stop_trim, list)
+                        else request.no_stop_trim[i]
                     ),
                 }
             )
diff --git a/python/sglang/srt/openai_api/protocol.py b/python/sglang/srt/openai_api/protocol.py
index 4b382240a..1da27af28 100644
--- a/python/sglang/srt/openai_api/protocol.py
+++ b/python/sglang/srt/openai_api/protocol.py
@@ -174,7 +174,7 @@ class CompletionRequest(BaseModel):
     min_tokens: int = 0
     repetition_penalty: Optional[float] = 1.0
     stop_token_ids: Optional[List[int]] = Field(default_factory=list)
-    no_eos_trim: Union[bool, List[bool]] = False
+    no_stop_trim: Union[bool, List[bool]] = False
 
 
 class CompletionResponseChoice(BaseModel):
diff --git a/python/sglang/srt/sampling/sampling_params.py b/python/sglang/srt/sampling/sampling_params.py
index 6e497ea7b..55ec00fc1 100644
--- a/python/sglang/srt/sampling/sampling_params.py
+++ b/python/sglang/srt/sampling/sampling_params.py
@@ -40,7 +40,7 @@ class SamplingParams:
         regex: Optional[str] = None,
         n: int = 1,
         json_schema: Optional[str] = None,
-        no_eos_trim: bool = False,
+        no_stop_trim: bool = False,
     ) -> None:
         self.temperature = temperature
         self.top_p = top_p
@@ -61,7 +61,7 @@ class SamplingParams:
         self.regex = regex
         self.n = n
         self.json_schema = json_schema
-        self.no_eos_trim = no_eos_trim
+        self.no_stop_trim = no_stop_trim
 
         # Process some special cases
         if self.temperature < _SAMPLING_EPS: