[Refactor] [6/N] to simplify the vLLM openai chat_completion serving architecture (#32240)

Signed-off-by: chaunceyjiang <chaunceyjiang@gmail.com>

vllm/entrypoints/serve/tokenize/serving.py

@@ -9,18 +9,20 @@ from fastapi import Request
 from vllm.engine.protocol import EngineClient
 from vllm.entrypoints.chat_utils import ChatTemplateContentFormatOption
 from vllm.entrypoints.logger import RequestLogger
-from vllm.entrypoints.openai.protocol import (
+from vllm.entrypoints.openai.engine.protocol import (
+    ErrorResponse,
+)
+from vllm.entrypoints.openai.engine.serving import OpenAIServing
+from vllm.entrypoints.openai.serving_models import OpenAIServingModels
+from vllm.entrypoints.renderer import RenderConfig
+from vllm.entrypoints.serve.tokenize.protocol import (
     DetokenizeRequest,
     DetokenizeResponse,
-    ErrorResponse,
     TokenizeChatRequest,
     TokenizeRequest,
     TokenizeResponse,
     TokenizerInfoResponse,
 )
-from vllm.entrypoints.openai.serving_engine import OpenAIServing
-from vllm.entrypoints.openai.serving_models import OpenAIServingModels
-from vllm.entrypoints.renderer import RenderConfig
 from vllm.inputs import TokensPrompt
 from vllm.logger import init_logger
 from vllm.tokenizers import TokenizerLike

vllm/entrypoints/utils.py

@@ -22,9 +22,11 @@ from vllm.entrypoints.chat_utils import (
     resolve_hf_chat_template,
     resolve_mistral_chat_template,
 )
-from vllm.entrypoints.openai.cli_args import make_arg_parser
-from vllm.entrypoints.openai.protocol import (
+from vllm.entrypoints.openai.chat_completion.protocol import (
     ChatCompletionRequest,
+)
+from vllm.entrypoints.openai.cli_args import make_arg_parser
+from vllm.entrypoints.openai.engine.protocol import (
     CompletionRequest,
     StreamOptions,
 )

vllm/reasoning/abs_reasoning_parsers.py

@@ -14,8 +14,10 @@ from vllm.utils.collection_utils import is_list_of
 from vllm.utils.import_utils import import_from_path
 
 if TYPE_CHECKING:
-    from vllm.entrypoints.openai.protocol import (
+    from vllm.entrypoints.openai.chat_completion.protocol import (
         ChatCompletionRequest,
+    )
+    from vllm.entrypoints.openai.engine.protocol import (
         DeltaMessage,
         ResponsesRequest,
     )

vllm/reasoning/basic_parsers.py

@@ -5,13 +5,15 @@ from abc import abstractmethod
 from collections.abc import Sequence
 from typing import TYPE_CHECKING, Any
 
-from vllm.entrypoints.openai.protocol import DeltaMessage
+from vllm.entrypoints.openai.engine.protocol import DeltaMessage
 from vllm.reasoning.abs_reasoning_parsers import ReasoningParser
 from vllm.tokenizers import TokenizerLike
 
 if TYPE_CHECKING:
-    from vllm.entrypoints.openai.protocol import (
+    from vllm.entrypoints.openai.chat_completion.protocol import (
         ChatCompletionRequest,
+    )
+    from vllm.entrypoints.openai.engine.protocol import (
         ResponsesRequest,
     )
 else:

vllm/reasoning/deepseek_r1_reasoning_parser.py

@@ -3,7 +3,7 @@
 
 from collections.abc import Sequence
 
-from vllm.entrypoints.openai.protocol import DeltaMessage
+from vllm.entrypoints.openai.engine.protocol import DeltaMessage
 from vllm.reasoning.basic_parsers import BaseThinkingReasoningParser
 
 

vllm/reasoning/deepseek_v3_reasoning_parser.py

@@ -5,7 +5,10 @@ from collections.abc import Sequence
 
 from transformers import PreTrainedTokenizerBase
 
-from vllm.entrypoints.openai.protocol import ChatCompletionRequest, DeltaMessage
+from vllm.entrypoints.openai.chat_completion.protocol import (
+    ChatCompletionRequest,
+)
+from vllm.entrypoints.openai.engine.protocol import DeltaMessage
 from vllm.logger import init_logger
 from vllm.reasoning import ReasoningParser
 from vllm.reasoning.deepseek_r1_reasoning_parser import DeepSeekR1ReasoningParser

vllm/reasoning/ernie45_reasoning_parser.py

@@ -5,7 +5,10 @@ from collections.abc import Sequence
 
 from transformers import PreTrainedTokenizerBase
 
-from vllm.entrypoints.openai.protocol import ChatCompletionRequest, DeltaMessage
+from vllm.entrypoints.openai.chat_completion.protocol import (
+    ChatCompletionRequest,
+)
+from vllm.entrypoints.openai.engine.protocol import DeltaMessage
 from vllm.logger import init_logger
 from vllm.reasoning.basic_parsers import BaseThinkingReasoningParser
 

vllm/reasoning/gptoss_reasoning_parser.py

@@ -5,8 +5,11 @@ from collections.abc import Sequence
 
 from transformers import PreTrainedTokenizerBase
 
+from vllm.entrypoints.openai.chat_completion.protocol import (
+    ChatCompletionRequest,
+)
+from vllm.entrypoints.openai.engine.protocol import DeltaMessage
 from vllm.entrypoints.openai.parser.harmony_utils import parse_chat_output
-from vllm.entrypoints.openai.protocol import ChatCompletionRequest, DeltaMessage
 from vllm.entrypoints.tool_server import ToolServer
 from vllm.logger import init_logger
 from vllm.reasoning import ReasoningParser

vllm/reasoning/granite_reasoning_parser.py

@@ -6,7 +6,10 @@ from collections.abc import Sequence
 import regex as re
 from transformers import PreTrainedTokenizerBase
 
-from vllm.entrypoints.openai.protocol import ChatCompletionRequest, DeltaMessage
+from vllm.entrypoints.openai.chat_completion.protocol import (
+    ChatCompletionRequest,
+)
+from vllm.entrypoints.openai.engine.protocol import DeltaMessage
 from vllm.logger import init_logger
 from vllm.reasoning import ReasoningParser
 

vllm/reasoning/holo2_reasoning_parser.py

@@ -3,7 +3,10 @@
 
 from collections.abc import Sequence
 
-from vllm.entrypoints.openai.protocol import ChatCompletionRequest, DeltaMessage
+from vllm.entrypoints.openai.chat_completion.protocol import (
+    ChatCompletionRequest,
+)
+from vllm.entrypoints.openai.engine.protocol import DeltaMessage
 from vllm.logger import init_logger
 from vllm.reasoning import (
     ReasoningParser,

vllm/reasoning/hunyuan_a13b_reasoning_parser.py

@@ -6,7 +6,10 @@ from collections.abc import Sequence
 import regex as re
 from transformers import PreTrainedTokenizerBase
 
-from vllm.entrypoints.openai.protocol import ChatCompletionRequest, DeltaMessage
+from vllm.entrypoints.openai.chat_completion.protocol import (
+    ChatCompletionRequest,
+)
+from vllm.entrypoints.openai.engine.protocol import DeltaMessage
 from vllm.logger import init_logger
 from vllm.reasoning import ReasoningParser
 

vllm/reasoning/identity_reasoning_parser.py

@@ -5,7 +5,10 @@ from collections.abc import Sequence
 
 from transformers import PreTrainedTokenizerBase
 
-from vllm.entrypoints.openai.protocol import ChatCompletionRequest, DeltaMessage
+from vllm.entrypoints.openai.chat_completion.protocol import (
+    ChatCompletionRequest,
+)
+from vllm.entrypoints.openai.engine.protocol import DeltaMessage
 from vllm.logger import init_logger
 from vllm.reasoning import ReasoningParser
 

vllm/reasoning/minimax_m2_reasoning_parser.py

@@ -3,8 +3,10 @@
 
 from collections.abc import Sequence
 
-from vllm.entrypoints.openai.protocol import (
+from vllm.entrypoints.openai.chat_completion.protocol import (
     ChatCompletionRequest,
+)
+from vllm.entrypoints.openai.engine.protocol import (
     DeltaMessage,
     ResponsesRequest,
 )

vllm/reasoning/mistral_reasoning_parser.py

@@ -3,8 +3,10 @@
 
 from functools import cached_property
 
-from vllm.entrypoints.openai.protocol import (
+from vllm.entrypoints.openai.chat_completion.protocol import (
     ChatCompletionRequest,
+)
+from vllm.entrypoints.openai.engine.protocol import (
     ResponsesRequest,
 )
 from vllm.logger import init_logger

vllm/reasoning/olmo3_reasoning_parser.py

@@ -10,9 +10,10 @@ import regex as re
 
 if TYPE_CHECKING:
     from vllm.tokenizers import TokenizerLike
-
-from vllm.entrypoints.openai.protocol import (
+from vllm.entrypoints.openai.chat_completion.protocol import (
     ChatCompletionRequest,
+)
+from vllm.entrypoints.openai.engine.protocol import (
     DeltaMessage,
     ResponsesRequest,
 )

vllm/reasoning/qwen3_reasoning_parser.py

 # SPDX-License-Identifier: Apache-2.0
 # SPDX-FileCopyrightText: Copyright contributors to the vLLM project
 
-
-from vllm.entrypoints.openai.protocol import ChatCompletionRequest, ResponsesRequest
+from vllm.entrypoints.openai.chat_completion.protocol import (
+    ChatCompletionRequest,
+)
+from vllm.entrypoints.openai.engine.protocol import ResponsesRequest
 from vllm.reasoning.basic_parsers import BaseThinkingReasoningParser
 
 

vllm/reasoning/step3_reasoning_parser.py

@@ -6,7 +6,10 @@ from collections.abc import Sequence
 import regex as re
 from transformers import PreTrainedTokenizerBase
 
-from vllm.entrypoints.openai.protocol import ChatCompletionRequest, DeltaMessage
+from vllm.entrypoints.openai.chat_completion.protocol import (
+    ChatCompletionRequest,
+)
+from vllm.entrypoints.openai.engine.protocol import DeltaMessage
 from vllm.logger import init_logger
 from vllm.reasoning import ReasoningParser
 

vllm/tokenizers/mistral.py

@@ -19,7 +19,7 @@ from mistral_common.tokens.tokenizers.sentencepiece import (
 from mistral_common.tokens.tokenizers.tekken import Tekkenizer
 
 from vllm.entrypoints.chat_utils import ChatCompletionMessageParam
-from vllm.entrypoints.openai.protocol import ChatCompletionRequest
+from vllm.entrypoints.openai.chat_completion.protocol import ChatCompletionRequest
 from vllm.logger import init_logger
 
 from .protocol import TokenizerLike

vllm/tool_parsers/abstract_tool_parser.py

@@ -10,8 +10,8 @@ from openai.types.responses.response_format_text_json_schema_config import (
     ResponseFormatTextJSONSchemaConfig,
 )
 
-from vllm.entrypoints.openai.protocol import (
-    ChatCompletionRequest,
+from vllm.entrypoints.openai.chat_completion.protocol import ChatCompletionRequest
+from vllm.entrypoints.openai.engine.protocol import (
     DeltaMessage,
     ExtractedToolCallInformation,
     ResponsesRequest,

vllm/tool_parsers/deepseekv31_tool_parser.py

@@ -6,8 +6,10 @@ from collections.abc import Sequence
 import regex as re
 
 from vllm.entrypoints.chat_utils import make_tool_call_id
-from vllm.entrypoints.openai.protocol import (
+from vllm.entrypoints.openai.chat_completion.protocol import (
     ChatCompletionRequest,
+)
+from vllm.entrypoints.openai.engine.protocol import (
     DeltaFunctionCall,
     DeltaMessage,
     DeltaToolCall,