feat: add MiniMax as first-class LLM provider (chat + embedding)

README.md

@@ -123,7 +123,7 @@ Click [Github Latest Release](https://github.com/volcengine/MineContext/releases
 
 After the application launches, please follow the prompts to enter your API key. (Note: On the first run, the application needs to install the backend environment, which may take about two minutes).
 
-We currently support services from Doubao, OpenAI, and custom models. This includes any **local models** or **third-party model** services that are compatible with the OpenAI API format.
+We currently support services from Doubao, OpenAI, MiniMax, and custom models. This includes any **local models** or **third-party model** services that are compatible with the OpenAI API format.
 
 We recommend using [LMStudio](https://lmstudio.ai/) to run local models. It provides a simple interface and powerful features to help you quickly deploy and manage them.
 
@@ -326,7 +326,7 @@ opencontext/
 
 6. **LLM Integration** (`llm/`)
 
-   - Support for multiple LLM providers (OpenAI, Doubao)
+   - Support for multiple LLM providers (OpenAI, Doubao, MiniMax)
    - VLM (Vision-Language Model) integration
    - Embedding generation services
 
@@ -359,12 +359,12 @@ server:
   debug: false
 
 embedding_model:
-  provider: doubao # options: openai, doubao
+  provider: doubao # options: openai, doubao, minimax
   api_key: your-api-key
   model: doubao-embedding-vision-250615
 
 vlm_model:
-  provider: doubao # options: openai, doubao
+  provider: doubao # options: openai, doubao, minimax
   api_key: your-api-key
   model: doubao-seed-1-6-flash-250828
 

README_zh.md

@@ -121,7 +121,7 @@ MineContext 非常注重用户隐私，所有数据都默认保存在本地如
 
 ## 2. 输入您的 API 密钥
 
-应用程序启动后（首次运行时需要安装后端环境，约需等待两分钟），请根据引导输入您的 API 密钥。目前我们支持豆包、OpenAI 以及自定义模型服务，包括任何兼容 OpenAI API 格式的**本地模型**或**第三方模型**服务。
+应用程序启动后（首次运行时需要安装后端环境，约需等待两分钟），请根据引导输入您的 API 密钥。目前我们支持豆包、OpenAI、MiniMax 以及自定义模型服务，包括任何兼容 OpenAI API 格式的**本地模型**或**第三方模型**服务。
 我们推荐使用 [LMStudio](https://lmstudio.ai/) 来运行本地模型，它提供了简单的界面和强大的功能，能够帮助您快速部署和管理本地模型。
 
 **综合成本和性能，我们推荐使用豆包模型**，豆包模型的 API-Key 可以在 [API 管理界面](https://console.volcengine.com/ark/region:ark+cn-beijing/apiKey) 生成。
@@ -325,7 +325,7 @@ opencontext/
 
 6. **LLM 集成** (`llm/`)
 
-   - 支持多个 LLM 提供商（OpenAI、豆包）
+   - 支持多个 LLM 提供商（OpenAI、豆包、MiniMax）
    - VLM（视觉-语言模型）集成
    - 嵌入生成服务
 
@@ -358,12 +358,12 @@ server:
   debug: false
 
 embedding_model:
-  provider: doubao # 选项：openai, doubao
+  provider: doubao # 选项：openai, doubao, minimax
   api_key: your-api-key
   model: doubao-embedding-vision-250615
 
 vlm_model:
-  provider: doubao # 选项：openai, doubao
+  provider: doubao # 选项：openai, doubao, minimax
   api_key: your-api-key
   model: doubao-seed-1-6-flash-250828
 

frontend/src/renderer/src/pages/settings/constants.tsx

@@ -4,21 +4,25 @@
 import { ReactNode } from 'react'
 import openAI from '../../assets/images/settings/OpenAI.png'
 import doubao from '../../assets/images/settings/doubao.png'
+import minimax from '../../assets/images/settings/minimax.svg'
 import custom from '../../assets/images/settings/custom.svg'
 
 export enum ModelTypeList {
   Doubao = 'doubao',
   OpenAI = 'openai',
+  MiniMax = 'minimax',
   Custom = 'custom'
 }
 
 export enum embeddingModels {
   DoubaoEmbeddingModelId = 'doubao-embedding-vision-250615',
-  OpenAIEmbeddingModelId = 'text-embedding-3-large'
+  OpenAIEmbeddingModelId = 'text-embedding-3-large',
+  MiniMaxEmbeddingModelId = 'embo-01'
 }
 export enum BaseUrl {
   DoubaoUrl = 'https://ark.cn-beijing.volces.com/api/v3',
-  OpenAIUrl = 'https://api.openai.com/v1'
+  OpenAIUrl = 'https://api.openai.com/v1',
+  MiniMaxUrl = 'https://api.minimax.io/v1'
 }
 export interface OptionInfo {
   value: string
@@ -70,6 +74,21 @@ export const ModelInfoList = [
       }
     ]
   },
+  {
+    icon: <img src={minimax} className="!max-w-none w-[24px] h-[24px]" />,
+    key: 'MiniMax',
+    value: 'minimax',
+    option: [
+      {
+        value: 'MiniMax-M2.7',
+        label: 'MiniMax-M2.7'
+      },
+      {
+        value: 'MiniMax-M2.7-highspeed',
+        label: 'MiniMax-M2.7-highspeed'
+      }
+    ]
+  },
   {
     icon: <img src={custom} className="!max-w-none w-[18px] h-[18px]" />,
     key: 'Custom',

frontend/src/renderer/src/pages/settings/settings.tsx

@@ -157,14 +157,19 @@ const StandardFormItems: FC<StandardFormItemsProps> = (props) => {
             You can get the API Key Here:
             <Button
               onClick={() => {
-                const url =
-                  modelPlatform === ModelTypeList.Doubao
-                    ? 'https://www.volcengine.com/docs/82379/1541594'
-                    : 'https://platform.openai.com/settings/organization/api-keys'
-                window.open(`${url}`)
+                const urlMap: Record<string, string> = {
+                  [ModelTypeList.Doubao]: 'https://www.volcengine.com/docs/82379/1541594',
+                  [ModelTypeList.MiniMax]: 'https://platform.minimaxi.com/user-center/basic-information/interface-key',
+                  [ModelTypeList.OpenAI]: 'https://platform.openai.com/settings/organization/api-keys'
+                }
+                window.open(urlMap[modelPlatform] || urlMap[ModelTypeList.OpenAI])
               }}
               type="text">
-              {modelPlatform === ModelTypeList.Doubao ? 'Get Doubao API Key' : 'Get OpenAI API Key'}
+              {modelPlatform === ModelTypeList.Doubao
+                ? 'Get Doubao API Key'
+                : modelPlatform === ModelTypeList.MiniMax
+                  ? 'Get MiniMax API Key'
+                  : 'Get OpenAI API Key'}
             </Button>
           </div>
         }
@@ -246,15 +251,22 @@ const Settings: FC<SettingsProps> = (props) => {
       const formatData = Object.fromEntries(
         Object.entries(data).map(([key, value]) => [key.replace(`${values.modelPlatform}-`, ''), value])
       )
+      const getBaseUrl = (platform: string) => {
+        if (platform === ModelTypeList.Doubao) return BaseUrl.DoubaoUrl
+        if (platform === ModelTypeList.MiniMax) return BaseUrl.MiniMaxUrl
+        return BaseUrl.OpenAIUrl
+      }
+      const getEmbeddingModelId = (platform: string) => {
+        if (platform === ModelTypeList.Doubao) return embeddingModels.DoubaoEmbeddingModelId
+        if (platform === ModelTypeList.MiniMax) return embeddingModels.MiniMaxEmbeddingModelId
+        return embeddingModels.OpenAIEmbeddingModelId
+      }
       const params = isCustom
         ? formatData
         : {
             ...formatData,
-            baseUrl: values.modelPlatform === ModelTypeList.Doubao ? BaseUrl.DoubaoUrl : BaseUrl.OpenAIUrl,
-            embeddingModelId:
-              values.modelPlatform === ModelTypeList.Doubao
-                ? embeddingModels.DoubaoEmbeddingModelId
-                : embeddingModels.OpenAIEmbeddingModelId
+            baseUrl: getBaseUrl(values.modelPlatform),
+            embeddingModelId: getEmbeddingModelId(values.modelPlatform)
           }
 
       updateModelSettings(params as unknown as ModelConfigProps)
@@ -299,7 +311,8 @@ const Settings: FC<SettingsProps> = (props) => {
               initialValues={{
                 modelPlatform: ModelTypeList.Doubao,
                 [`${ModelTypeList.Doubao}-modelId`]: 'doubao-seed-1-6-flash-250828',
-                [`${ModelTypeList.OpenAI}-modelId`]: 'gpt-5-nano'
+                [`${ModelTypeList.OpenAI}-modelId`]: 'gpt-5-nano',
+                [`${ModelTypeList.MiniMax}-modelId`]: 'MiniMax-M2.7'
               }}>
               <FormItem label="Model platform" field={'modelPlatform'} requiredSymbol={false}>
                 <ModelRadio />
@@ -315,6 +328,8 @@ const Settings: FC<SettingsProps> = (props) => {
                     return <StandardFormItems modelPlatform={modelPlatform} prefix={ModelTypeList.Doubao} />
                   } else if (modelPlatform === ModelTypeList.OpenAI) {
                     return <StandardFormItems modelPlatform={modelPlatform} prefix={ModelTypeList.OpenAI} />
+                  } else if (modelPlatform === ModelTypeList.MiniMax) {
+                    return <StandardFormItems modelPlatform={modelPlatform} prefix={ModelTypeList.MiniMax} />
                   } else {
                     return null
                   }

opencontext/llm/llm_client.py

@@ -23,6 +23,7 @@
 class LLMProvider(Enum):
     OPENAI = "openai"
     DOUBAO = "doubao"
+    MINIMAX = "minimax"
 
 
 class LLMType(Enum):
@@ -264,9 +265,42 @@ async def _openai_chat_completion_stream_async(self, messages: List[Dict[str, An
             logger.error(f"OpenAI API async stream error: {e}")
             raise
 
+    def _minimax_embedding(self, text: str, **kwargs) -> List[float]:
+        """Request embedding from MiniMax's native embedding API (embo-01).
+
+        MiniMax uses a non-OpenAI-compatible format:
+        - Request: {"model": "embo-01", "texts": [...], "type": "db"|"query"}
+        - Response: {"vectors": [[...]], "total_tokens": N, "base_resp": {...}}
+        """
+        import httpx
+
+        url = f"{self.base_url.rstrip('/')}/embeddings"
+        headers = {
+            "Authorization": f"Bearer {self.api_key}",
+            "Content-Type": "application/json",
+        }
+        payload = {
+            "model": self.model,
+            "texts": [text],
+            "type": kwargs.get("embedding_type", "db"),
+        }
+        resp = httpx.post(url, json=payload, headers=headers, timeout=self.timeout)
+        resp.raise_for_status()
+        data = resp.json()
+        base_resp = data.get("base_resp", {})
+        if base_resp.get("status_code", 0) != 0:
+            raise ValueError(f"MiniMax embedding error: {base_resp.get('status_msg', 'unknown')}")
+        vectors = data.get("vectors", [])
+        if not vectors:
+            raise ValueError("MiniMax embedding returned empty vectors")
+        return vectors[0]
+
     def _request_embedding(self, text: str, **kwargs) -> List[float]:
         try:
-            if self.provider != LLMProvider.DOUBAO.value:
+            response = None
+            if self.provider == LLMProvider.MINIMAX.value:
+                embedding = self._minimax_embedding(text)
+            elif self.provider != LLMProvider.DOUBAO.value:
                 response = self.client.embeddings.create(model=self.model, input=[text])
                 embedding = response.data[0].embedding
             else:
@@ -276,7 +310,7 @@ def _request_embedding(self, text: str, **kwargs) -> List[float]:
                 embedding = response.data.embedding
 
             # Record token usage
-            if hasattr(response, "usage") and response.usage:
+            if response and hasattr(response, "usage") and response.usage:
                 try:
                     from opencontext.monitoring import record_token_usage
 
@@ -311,9 +345,38 @@ def _request_embedding(self, text: str, **kwargs) -> List[float]:
             logger.error(f"OpenAI API error during embedding: {e}")
             raise
 
+    async def _minimax_embedding_async(self, text: str, **kwargs) -> List[float]:
+        """Async request embedding from MiniMax's native embedding API."""
+        import httpx
+
+        url = f"{self.base_url.rstrip('/')}/embeddings"
+        headers = {
+            "Authorization": f"Bearer {self.api_key}",
+            "Content-Type": "application/json",
+        }
+        payload = {
+            "model": self.model,
+            "texts": [text],
+            "type": kwargs.get("embedding_type", "db"),
+        }
+        async with httpx.AsyncClient(timeout=self.timeout) as client:
+            resp = await client.post(url, json=payload, headers=headers)
+            resp.raise_for_status()
+        data = resp.json()
+        base_resp = data.get("base_resp", {})
+        if base_resp.get("status_code", 0) != 0:
+            raise ValueError(f"MiniMax embedding error: {base_resp.get('status_msg', 'unknown')}")
+        vectors = data.get("vectors", [])
+        if not vectors:
+            raise ValueError("MiniMax embedding returned empty vectors")
+        return vectors[0]
+
     async def _request_embedding_async(self, text: str, **kwargs) -> List[float]:
         try:
-            if self.provider == LLMProvider.DOUBAO.value:
+            response = None
+            if self.provider == LLMProvider.MINIMAX.value:
+                embedding = await self._minimax_embedding_async(text, **kwargs)
+            elif self.provider == LLMProvider.DOUBAO.value:
                 # Only ark has multimodal_embeddings
                 response = self.client.multimodal_embeddings.create(
                     model=self.model, input=[{"type": "text", "text": text}]
@@ -324,7 +387,7 @@ async def _request_embedding_async(self, text: str, **kwargs) -> List[float]:
                 embedding = response.data[0].embedding
 
             # Record token usage
-            if hasattr(response, "usage") and response.usage:
+            if response and hasattr(response, "usage") and response.usage:
                 try:
                     from opencontext.monitoring import record_token_usage
 
@@ -406,7 +469,17 @@ def _extract_error_summary(error: Any) -> str:
                 if code in error_msg:
                     return msg
 
-            # 2. Check for OpenAI specific errors
+            # 2. Check for MiniMax specific errors
+            minimax_errors = {
+                "invalid_api_key": "Invalid MiniMax API key.",
+                "insufficient_balance": "Insufficient MiniMax account balance.",
+            }
+
+            for code, msg in minimax_errors.items():
+                if code in error_msg:
+                    return msg
+
+            # 3. Check for OpenAI specific errors
             openai_errors = {
                 "insufficient_quota": "Insufficient quota. Check your plan and billing details.",
                 "invalid_api_key": "Invalid API key provided.",
@@ -477,7 +550,13 @@ def _extract_error_summary(error: Any) -> str:
 
             elif self.llm_type == LLMType.EMBEDDING:
                 # Test with a simple text
-                if self.provider == LLMProvider.DOUBAO.value:
+                if self.provider == LLMProvider.MINIMAX.value:
+                    embedding = self._minimax_embedding("test")
+                    if embedding and len(embedding) > 0:
+                        return True, "Embedding model validation successful"
+                    else:
+                        return False, "Embedding model returned empty response"
+                elif self.provider == LLMProvider.DOUBAO.value:
                     response = self.client.multimodal_embeddings.create(
                         model=self.model, input=[{"type": "text", "text": "test"}]
                     )

pyproject.toml

@@ -44,13 +44,16 @@ dependencies = [
     "python-multipart",
     "playwright",
     "volcengine-python-sdk[ark]",
+    "httpx",
 ]
 
 [project.optional-dependencies]
 dev = [
     "black>=24.8.0",
     "isort>=5.13.0",
     "pre-commit>=3.6.0",
+    "pytest>=7.0.0",
+    "pytest-asyncio>=0.21.0",
 ]
 
 [project.scripts]
@@ -90,6 +93,12 @@ extend-exclude = '''
 )/
 '''
 
+# pytest - test configuration
+[tool.pytest.ini_options]
+markers = [
+    "integration: marks tests that require a real MiniMax API key (deselect with '-m \"not integration\"')",
+]
+
 # isort - import 语句排序工具
 [tool.isort]
 profile = "black"