Refa: add MiniMax-M2 and remove deprecated MiniMax models (#11642)

### What problem does this PR solve? Add MiniMax-M2 and remove deprecated models. ### Type of change - [x] New Feature (non-breaking change which adds functionality) - [x] Refactoring
2025-12-06 07:19:03 +08:00 · 2025-12-02 14:43:44 +08:00
parent 519f03097e
commit a713f54732
15 changed files with 62 additions and 112 deletions
--- a/conf/llm_factories.json
+++ b/conf/llm_factories.json
@@ -1226,39 +1226,14 @@
        {
            "name": "MiniMax",
            "logo": "",
-            "tags": "LLM,TEXT EMBEDDING",
+            "tags": "LLM",
            "status": "1",
            "rank": "810",
            "llm": [
                {
-                    "llm_name": "abab6.5-chat",
-                    "tags": "LLM,CHAT,8k",
-                    "max_tokens": 8192,
-                    "model_type": "chat"
-                },
-                {
-                    "llm_name": "abab6.5s-chat",
-                    "tags": "LLM,CHAT,245k",
-                    "max_tokens": 245760,
-                    "model_type": "chat",
-                    "is_tools": true
-                },
-                {
-                    "llm_name": "abab6.5t-chat",
-                    "tags": "LLM,CHAT,8k",
-                    "max_tokens": 8192,
-                    "model_type": "chat"
-                },
-                {
-                    "llm_name": "abab6.5g-chat",
-                    "tags": "LLM,CHAT,8k",
-                    "max_tokens": 8192,
-                    "model_type": "chat"
-                },
-                {
-                    "llm_name": "abab5.5s-chat",
-                    "tags": "LLM,CHAT,8k",
-                    "max_tokens": 8192,
+                    "llm_name": "MiniMax-M2",
+                    "tags": "LLM,CHAT,200k",
+                    "max_tokens": 200000,
                    "model_type": "chat"
                }
            ]
--- a/rag/llm/init.py
+++ b/rag/llm/init.py
@@ -51,6 +51,7 @@ class SupportedLiteLLMProvider(StrEnum):
    AI_302 = "302.AI"
    JiekouAI = "Jiekou.AI"
    ZHIPU_AI = "ZHIPU-AI"
+    MiniMax = "MiniMax"


 FACTORY_DEFAULT_BASE_URL = {
@@ -73,6 +74,7 @@ FACTORY_DEFAULT_BASE_URL = {
    SupportedLiteLLMProvider.Anthropic: "https://api.anthropic.com/",
    SupportedLiteLLMProvider.JiekouAI: "https://api.jiekou.ai/openai",
    SupportedLiteLLMProvider.ZHIPU_AI: "https://open.bigmodel.cn/api/paas/v4",
+    SupportedLiteLLMProvider.MiniMax: "https://api.minimaxi.com/v1",
 }


@@ -105,6 +107,7 @@ LITELLM_PROVIDER_PREFIX = {
    SupportedLiteLLMProvider.AI_302: "openai/",
    SupportedLiteLLMProvider.JiekouAI: "openai/",
    SupportedLiteLLMProvider.ZHIPU_AI: "openai/",
+    SupportedLiteLLMProvider.MiniMax: "openai/",
 }

 ChatModel = globals().get("ChatModel", {})
--- a/rag/llm/chat_model.py
+++ b/rag/llm/chat_model.py
@@ -28,7 +28,6 @@ from urllib.parse import urljoin
 import json_repair
 import litellm
 import openai
-import requests
 from openai import AsyncOpenAI, OpenAI
 from openai.lib.azure import AzureOpenAI
 from strenum import StrEnum
@@ -1015,86 +1014,6 @@ class VolcEngineChat(Base):
        super().__init__(ark_api_key, model_name, base_url, **kwargs)


-class MiniMaxChat(Base):
-    _FACTORY_NAME = "MiniMax"
-
-    def __init__(self, key, model_name, base_url="https://api.minimax.chat/v1/text/chatcompletion_v2", **kwargs):
-        super().__init__(key, model_name, base_url=base_url, **kwargs)
-
-        if not base_url:
-            base_url = "https://api.minimax.chat/v1/text/chatcompletion_v2"
-        self.base_url = base_url
-        self.model_name = model_name
-        self.api_key = key
-
-    def _clean_conf(self, gen_conf):
-        for k in list(gen_conf.keys()):
-            if k not in ["temperature", "top_p", "max_tokens"]:
-                del gen_conf[k]
-        return gen_conf
-
-    def _chat(self, history, gen_conf):
-        headers = {
-            "Authorization": f"Bearer {self.api_key}",
-            "Content-Type": "application/json",
-        }
-        payload = json.dumps({"model": self.model_name, "messages": history, **gen_conf})
-        response = requests.request("POST", url=self.base_url, headers=headers, data=payload)
-        response = response.json()
-        ans = response["choices"][0]["message"]["content"].strip()
-        if response["choices"][0]["finish_reason"] == "length":
-            if is_chinese(ans):
-                ans += LENGTH_NOTIFICATION_CN
-            else:
-                ans += LENGTH_NOTIFICATION_EN
-        return ans, total_token_count_from_response(response)
-
-    def chat_streamly(self, system, history, gen_conf):
-        if system and history and history[0].get("role") != "system":
-            history.insert(0, {"role": "system", "content": system})
-        for k in list(gen_conf.keys()):
-            if k not in ["temperature", "top_p", "max_tokens"]:
-                del gen_conf[k]
-        ans = ""
-        total_tokens = 0
-        try:
-            headers = {
-                "Authorization": f"Bearer {self.api_key}",
-                "Content-Type": "application/json",
-            }
-            payload = json.dumps(
-                {
-                    "model": self.model_name,
-                    "messages": history,
-                    "stream": True,
-                    **gen_conf,
-                }
-            )
-            response = requests.request(
-                "POST",
-                url=self.base_url,
-                headers=headers,
-                data=payload,
-            )
-            for resp in response.text.split("\n\n")[:-1]:
-                resp = json.loads(resp[6:])
-                text = ""
-                if "choices" in resp and "delta" in resp["choices"][0]:
-                    text = resp["choices"][0]["delta"]["content"]
-                ans = text
-                tol = total_token_count_from_response(resp)
-                if not tol:
-                    total_tokens += num_tokens_from_string(text)
-                else:
-                    total_tokens = tol
-                yield ans
-
-        except Exception as e:
-            yield ans + "\n**ERROR**: " + str(e)
-
-        yield total_tokens
-
-
 class MistralChat(Base):
    _FACTORY_NAME = "Mistral"

@@ -1642,6 +1561,7 @@ class LiteLLMBase(ABC):
        "302.AI",
        "Jiekou.AI",
        "ZHIPU-AI",
+        "MiniMax",
    ]

    def __init__(self, key, model_name, base_url=None, **kwargs):
--- a/web/src/locales/de.ts
+++ b/web/src/locales/de.ts
@@ -628,6 +628,10 @@ export default {
        'Für chinesische Benutzer ist keine Eingabe erforderlich oder verwenden Sie https://dashscope.aliyuncs.com/compatible-mode/v1. Für internationale Benutzer verwenden Sie https://dashscope-intl.aliyuncs.com/compatible-mode/v1',
      tongyiBaseUrlPlaceholder:
        '(Nur für internationale Benutzer, bitte Hinweis beachten)',
+      minimaxBaseUrlTip:
+        'Nur für internationale Nutzer: https://api.minimax.io/v1 verwenden.',
+      minimaxBaseUrlPlaceholder:
+        '(Nur für internationale Benutzer, https://api.minimax.io/v1 eintragen)',
      modify: 'Ändern',
      systemModelSettings: 'Standardmodelle festlegen',
      chatModel: 'Chat-Modell',
--- a/web/src/locales/en.ts
+++ b/web/src/locales/en.ts
@@ -858,6 +858,10 @@ Example: Virtual Hosted Style`,
      tongyiBaseUrlTip:
        'For Chinese users, no need to fill in or use https://dashscope.aliyuncs.com/compatible-mode/v1. For international users, use https://dashscope-intl.aliyuncs.com/compatible-mode/v1',
      tongyiBaseUrlPlaceholder: '(International users only, please see tip)',
+      minimaxBaseUrlTip:
+        'International users only: use https://api.minimax.io/v1',
+      minimaxBaseUrlPlaceholder:
+        '(International users only, fill in https://api.minimax.io/v1)',
      modify: 'Modify',
      systemModelSettings: 'Set default models',
      chatModel: 'LLM',
--- a/web/src/locales/es.ts
+++ b/web/src/locales/es.ts
@@ -344,6 +344,10 @@ export default {
        'Para usuarios chinos, no es necesario rellenar o usar https://dashscope.aliyuncs.com/compatible-mode/v1. Para usuarios internacionales, usar https://dashscope-intl.aliyuncs.com/compatible-mode/v1',
      tongyiBaseUrlPlaceholder:
        '(Solo para usuarios internacionales, por favor ver consejo)',
+      minimaxBaseUrlTip:
+        'Solo usuarios internacionales: utilice https://api.minimax.io/v1.',
+      minimaxBaseUrlPlaceholder:
+        '(Solo usuarios internacionales, ingrese https://api.minimax.io/v1)',
      modify: 'Modificar',
      systemModelSettings: 'Establecer modelos predeterminados',
      chatModel: 'Modelo de chat',
--- a/web/src/locales/fr.ts
+++ b/web/src/locales/fr.ts
@@ -526,6 +526,10 @@ export default {
        'Pour les utilisateurs chinois, pas besoin de remplir ou utiliser https://dashscope.aliyuncs.com/compatible-mode/v1. Pour les utilisateurs internationaux, utilisez https://dashscope-intl.aliyuncs.com/compatible-mode/v1',
      tongyiBaseUrlPlaceholder:
        "(Utilisateurs internationaux uniquement, veuillez consulter l'astuce)",
+      minimaxBaseUrlTip:
+        'Utilisateurs internationaux uniquement : utilisez https://api.minimax.io/v1.',
+      minimaxBaseUrlPlaceholder:
+        '(Utilisateurs internationaux uniquement, renseignez https://api.minimax.io/v1)',
      modify: 'Modifier',
      systemModelSettings: 'Définir les modèles par défaut',
      chatModel: 'Modèle de chat',
--- a/web/src/locales/id.ts
+++ b/web/src/locales/id.ts
@@ -516,6 +516,10 @@ export default {
        'Untuk pengguna Tiongkok, tidak perlu diisi atau gunakan https://dashscope.aliyuncs.com/compatible-mode/v1. Untuk pengguna internasional, gunakan https://dashscope-intl.aliyuncs.com/compatible-mode/v1',
      tongyiBaseUrlPlaceholder:
        '(Hanya untuk pengguna internasional, silakan lihat tip)',
+      minimaxBaseUrlTip:
+        'Hanya untuk pengguna internasional: gunakan https://api.minimax.io/v1.',
+      minimaxBaseUrlPlaceholder:
+        '(Hanya untuk pengguna internasional, isi https://api.minimax.io/v1)',
      modify: 'Ubah',
      systemModelSettings: 'Tetapkan model default',
      chatModel: 'Model Obrolan',
--- a/web/src/locales/ja.ts
+++ b/web/src/locales/ja.ts
@@ -557,6 +557,10 @@ export default {
      tongyiBaseUrlTip:
        '中国ユーザーの場合、記入不要または https://dashscope.aliyuncs.com/compatible-mode/v1 を使用してください。国際ユーザーは https://dashscope-intl.aliyuncs.com/compatible-mode/v1 を使用してください',
      tongyiBaseUrlPlaceholder: '（国際ユーザーのみ、ヒントをご覧ください）',
+      minimaxBaseUrlTip:
+        '国際ユーザーのみ：https://api.minimax.io/v1 を使用してください。',
+      minimaxBaseUrlPlaceholder:
+        '（国際ユーザーのみ、https://api.minimax.io/v1 を入力してください）',
      modify: '変更',
      systemModelSettings: 'デフォルトモデルを設定する',
      chatModel: 'チャットモデル',
--- a/web/src/locales/pt-br.ts
+++ b/web/src/locales/pt-br.ts
@@ -508,6 +508,10 @@ export default {
        'Para usuários chineses, não é necessário preencher ou usar https://dashscope.aliyuncs.com/compatible-mode/v1. Para usuários internacionais, use https://dashscope-intl.aliyuncs.com/compatible-mode/v1',
      tongyiBaseUrlPlaceholder:
        '(Apenas para usuários internacionais, consulte a dica)',
+      minimaxBaseUrlTip:
+        'Somente usuários internacionais: use https://api.minimax.io/v1.',
+      minimaxBaseUrlPlaceholder:
+        '(Somente para usuários internacionais, preencha https://api.minimax.io/v1)',
      modify: 'Modificar',
      systemModelSettings: 'Definir modelos padrão',
      chatModel: 'Modelo de chat',
--- a/web/src/locales/ru.ts
+++ b/web/src/locales/ru.ts
@@ -846,6 +846,10 @@ export default {
        'Для китайских пользователей не нужно заполнять или используйте https://dashscope.aliyuncs.com/compatible-mode/v1. Для международных пользователей используйте https://dashscope-intl.aliyuncs.com/compatible-mode/v1',
      tongyiBaseUrlPlaceholder:
        '(Только для международных пользователей, см. подсказку)',
+      minimaxBaseUrlTip:
+        'Только для международных пользователей: используйте https://api.minimax.io/v1.',
+      minimaxBaseUrlPlaceholder:
+        '(Только для международных пользователей, введите https://api.minimax.io/v1)',
      modify: 'Изменить',
      systemModelSettings: 'Установить модели по умолчанию',
      chatModel: 'LLM',
--- a/web/src/locales/vi.ts
+++ b/web/src/locales/vi.ts
@@ -558,6 +558,10 @@ export default {
      baseUrl: 'Base-Url',
      baseUrlTip:
        'Nếu khóa API của bạn từ OpenAI, chỉ cần bỏ qua nó. Bất kỳ nhà cung cấp trung gian nào khác sẽ cung cấp URL cơ sở này với khóa API.',
+      minimaxBaseUrlTip:
+        'Chỉ người dùng quốc tế: dùng https://api.minimax.io/v1.',
+      minimaxBaseUrlPlaceholder:
+        '(Chỉ dành cho người dùng quốc tế, điền https://api.minimax.io/v1)',
      modify: 'Sửa đổi',
      systemModelSettings: 'Đặt mô hình mặc định',
      chatModel: 'Mô hình trò chuyện',
--- a/web/src/locales/zh-traditional.ts
+++ b/web/src/locales/zh-traditional.ts
@@ -596,6 +596,8 @@ export default {
      tongyiBaseUrlTip:
        '中國用戶無需填寫或使用 https://dashscope.aliyuncs.com/compatible-mode/v1。國際用戶請使用 https://dashscope-intl.aliyuncs.com/compatible-mode/v1',
      tongyiBaseUrlPlaceholder: '（僅國際用戶，請參閱提示）',
+      minimaxBaseUrlTip: '僅國際用戶：使用 https://api.minimax.io/v1。',
+      minimaxBaseUrlPlaceholder: '（僅國際用戶填寫 https://api.minimax.io/v1）',
      modify: '修改',
      systemModelSettings: '設定預設模型',
      chatModel: '聊天模型',
--- a/web/src/locales/zh.ts
+++ b/web/src/locales/zh.ts
@@ -813,6 +813,8 @@ General：实体和关系提取提示来自 GitHub - microsoft/graphrag：基于
      tongyiBaseUrlTip:
        '对于中国用户，不需要填写或使用 https://dashscope.aliyuncs.com/compatible-mode/v1。对于国际用户，使用 https://dashscope-intl.aliyuncs.com/compatible-mode/v1。',
      tongyiBaseUrlPlaceholder: '(仅国际用户需要)',
+      minimaxBaseUrlTip: '仅国际用户：使用 https://api.minimax.io/v1。',
+      minimaxBaseUrlPlaceholder: '(仅国际用户填写 https://api.minimax.io/v1)',
      modify: '修改',
      systemModelSettings: '设置默认模型',
      chatModel: 'LLM',
--- a/web/src/pages/user-setting/setting-model/modal/api-key-modal/index.tsx
+++ b/web/src/pages/user-setting/setting-model/modal/api-key-modal/index.tsx
@@ -34,6 +34,7 @@ const modelsWithBaseUrl = [
  LLMFactory.OpenAI,
  LLMFactory.AzureOpenAI,
  LLMFactory.TongYiQianWen,
+  LLMFactory.MiniMax,
 ];

 const ApiKeyModal = ({
@@ -109,7 +110,16 @@ const ApiKeyModal = ({
              name="base_url"
              render={({ field }) => (
                <FormItem>
-                  <FormLabel className="text-sm font-medium text-text-primary">
+                  <FormLabel
+                    className="text-sm font-medium text-text-primary"
+                    tooltip={
+                      llmFactory === LLMFactory.MiniMax
+                        ? t('minimaxBaseUrlTip')
+                        : llmFactory === LLMFactory.TongYiQianWen
+                          ? t('tongyiBaseUrlTip')
+                          : t('baseUrlTip')
+                    }
+                  >
                    {t('baseUrl')}
                  </FormLabel>
                  <FormControl>
@@ -118,7 +128,9 @@ const ApiKeyModal = ({
                      placeholder={
                        llmFactory === LLMFactory.TongYiQianWen
                          ? t('tongyiBaseUrlPlaceholder')
-                          : 'https://api.openai.com/v1'
+                          : llmFactory === LLMFactory.MiniMax
+                            ? t('minimaxBaseUrlPlaceholder')
+                            : 'https://api.openai.com/v1'
                      }
                      onKeyDown={handleKeyDown}
                      className="w-full"