feat: 统一管理模型服务配置信息

app/config/ai_configs.py

@@ -0,0 +1,84 @@
+from app.config.env import env
+
+ai_configs = {
+  # /*---------------------------------------local-------------------------------------------*/
+  "local": {
+    "model": "deepseek-r1-distill-qwen-7b",
+    "url": "http://127.0.0.1:1234/v1/chat/completions",
+    "key": env.llm_key_local
+  },
+  "local_glm": {
+    "model": "glm-4-9b-0414",
+    "url": "http://127.0.0.1:1234/v1/chat/completions",
+    "key": env.llm_key_local
+  },
+  "local_embedding": {
+    "model": "text-embedding-text2vec-large-chinese",
+    "url": "http://127.0.0.1:1234/v1/embeddings",
+    "key": env.llm_key_local
+  },
+  # /*---------------------------------------deepseek-------------------------------------------*/
+  'deepseek-v3': {
+    'model': 'deepseek-chat',
+    'url': 'https://api.deepseek.com/chat/completions',
+    'key': env.llm_key_deepseek,
+  },
+  'deepseek-r1': {
+    'model': 'deepseek-reasoner',
+    'url': 'https://api.deepseek.com/chat/completions',
+    'key': env.llm_key_deepseek,
+  },
+  # /*---------------------------------------huoshan-------------------------------------------*/
+  'huoshan-deepseek-v3': {
+    'model': 'deepseek-v3-250324',
+    'url': 'https://ark.cn-beijing.volces.com/api/v3/chat/completions',
+    'key': env.llm_key_huoshan,
+  },
+  'huoshan-deepseek-r1': {
+    'model': 'deepseek-r1-distill-qwen-7b-250120',
+    'url': 'https://ark.cn-beijing.volces.com/api/v3/chat/completions',
+    'key': env.llm_key_huoshan,
+  },
+  'huoshan-doubao': {
+    'model': 'doubao-1-5-lite-32k-250115',
+    'url': 'https://ark.cn-beijing.volces.com/api/v3/chat/completions',
+    'key': env.llm_key_huoshan,
+  },
+  'huoshan-think-pro': {
+    'model': 'doubao-1-5-thinking-pro-250415',
+    'url': 'https://ark.cn-beijing.volces.com/api/v3/chat/completions',
+    'key': env.llm_key_huoshan,
+  },
+  'huoshan-doubao-seed': {
+    'model': 'doubao-seed-1-6-250615',
+    'url': 'https://ark.cn-beijing.volces.com/api/v3/chat/completions',
+    'key': env.llm_key_huoshan,
+  },
+  'doubao-vision-lite': {
+    'model': 'doubao-1.5-vision-lite-250315',
+    'url': 'https://ark.cn-beijing.volces.com/api/v3/chat/completions',
+    'key': env.llm_key_huoshan,
+  },
+
+  'huoshan-embedding-240715': {
+    "model": 'doubao-embedding-text-240715',
+    "url": 'https://ark.cn-beijing.volces.com/api/v3/embeddings',
+    'key': env.llm_key_huoshan,
+  },
+  # /*---------------------------------------bailian-------------------------------------------*/
+  'bailian-qwen-turbo': {
+    'model': 'qwen-turbo',
+    'url': 'https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions',
+    'key': env.llm_key_bailian,
+  },
+  'bailian-qwen3-max': {
+    'model': 'qwen3-max',
+    'url': 'https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions',
+    'key': env.llm_key_bailian,
+  },
+  'bailian-embedding': {
+    'model': 'text-embedding-v4',
+    'url': 'https://dashscope.aliyuncs.com/compatible-mode/v1/embeddings',
+    'key': env.llm_key_bailian,
+  },
+}

app/config/env.py

@@ -3,6 +3,11 @@ from pydantic import Field
 
 
 class EnvSettings(BaseSettings):
+  llm_key_local: str = Field(..., env="LLM_KEY_LOCAL")
+  llm_key_huoshan: str = Field(..., env="LLM_KEY_HUOSHAN")
+  llm_key_bailian: str = Field(..., env="LLM_KEY_BAILIAN")
+  llm_key_deepseek: str = Field(..., env="LLM_KEY_DEEPSEEK")
+
   server_port: int = Field(..., env="SERVER_PORT")
 
   class Config:

app/utils/llm_utils.py

@@ -0,0 +1,29 @@
+from langchain.chat_models import init_chat_model
+
+from app.config.ai_configs import ai_configs
+
+
+def create_llm(
+  platform_code='huoshan-doubao',
+  temperature=0.5,
+  disable_streaming=False,
+):
+  _ai_config = ai_configs.get(platform_code)
+
+  if _ai_config is None:
+    raise Exception('Unknown platform code', platform_code)
+
+  return init_chat_model(
+    # 必须填 openai（兼容协议）
+    model_provider="openai",
+    # 你的模型名
+    model=_ai_config.get('model'),
+    # 你的自定义 API URL
+    base_url=_ai_config.get('url').replace("chat/completions", ""),
+    # 本地模型随便填
+    api_key=_ai_config.get('key'),
+    # 随机性
+    temperature=temperature,
+    # 禁用流式响应
+    disable_streaming=disable_streaming,
+  )