增加extra_params功能

src/llm_models/model_client/base_client.py

@@ -83,6 +83,7 @@ class BaseClient:
         | None = None,
         async_response_parser: Callable[[ChatCompletion], tuple[APIResponse, tuple[int, int, int]]] | None = None,
         interrupt_flag: asyncio.Event | None = None,
+        extra_params: dict[str, Any] | None = None,
     ) -> APIResponse:
         """
         获取对话响应
@@ -103,6 +104,7 @@ class BaseClient:
         self,
         model_info: ModelInfo,
         embedding_input: str,
+        extra_params: dict[str, Any] | None = None,
     ) -> APIResponse:
         """
         获取文本嵌入

src/llm_models/model_client/gemini_client.py

@@ -1,3 +1,4 @@
+raise DeprecationWarning("Genimi Client is not fully available yet.")
 import asyncio
 import io
 from collections.abc import Iterable

src/llm_models/model_client/openai_client.py

@@ -396,6 +396,7 @@ class OpenaiClient(BaseClient):
             Callable[[ChatCompletion], tuple[APIResponse, Optional[tuple[int, int, int]]]]
         ] = None,
         interrupt_flag: asyncio.Event | None = None,
+        extra_params: dict[str, Any] | None = None,
     ) -> APIResponse:
         """
         获取对话响应
@@ -434,6 +435,7 @@ class OpenaiClient(BaseClient):
                         max_tokens=max_tokens,
                         stream=True,
                         response_format=NOT_GIVEN,
+                        extra_body=extra_params,
                     )
                 )
                 while not req_task.done():
@@ -455,6 +457,7 @@ class OpenaiClient(BaseClient):
                         max_tokens=max_tokens,
                         stream=False,
                         response_format=NOT_GIVEN,
+                        extra_body=extra_params,
                     )
                 )
                 while not req_task.done():
@@ -487,6 +490,7 @@ class OpenaiClient(BaseClient):
         self,
         model_info: ModelInfo,
         embedding_input: str,
+        extra_params: dict[str, Any] | None = None,
     ) -> APIResponse:
         """
         获取文本嵌入
@@ -498,6 +502,7 @@ class OpenaiClient(BaseClient):
             raw_response = await self.client.embeddings.create(
                 model=model_info.model_identifier,
                 input=embedding_input,
+                extra_body=extra_params,
             )
         except APIConnectionError as e:
             raise NetworkConnectionError() from e

template/model_config_template.toml

@@ -1,5 +1,5 @@
 [inner]
-version = "1.1.0"
+version = "1.1.1"
 
 # 配置文件版本号迭代规则同bot_config.toml
 
@@ -59,6 +59,8 @@ name = "qwen3-8b"
 api_provider = "SiliconFlow"
 price_in = 0
 price_out = 0
+[models.extra_params] # 可选的额外参数配置
+enable_thinking = false # 不启用思考
 
 [[models]]
 model_identifier = "Qwen/Qwen3-14B"
@@ -66,6 +68,8 @@ name = "qwen3-14b"
 api_provider = "SiliconFlow"
 price_in = 0.5
 price_out = 2.0
+[models.extra_params] # 可选的额外参数配置
+enable_thinking = false # 不启用思考
 
 [[models]]
 model_identifier = "Qwen/Qwen3-30B-A3B"
@@ -73,6 +77,8 @@ name = "qwen3-30b"
 api_provider = "SiliconFlow"
 price_in = 0.7
 price_out = 2.8
+[models.extra_params] # 可选的额外参数配置
+enable_thinking = false # 不启用思考
 
 [[models]]
 model_identifier = "Qwen/Qwen2.5-VL-72B-Instruct"