diff --git a/server/src/db/schema.ts b/server/src/db/schema.ts
index 2c48dfae..7b7a01d1 100644
--- a/server/src/db/schema.ts
+++ b/server/src/db/schema.ts
@@ -80,7 +80,8 @@ export function initializeSchema(db: Database.Database): void {
       custom_prompt TEXT,
       use_custom_prompt INTEGER DEFAULT 0,
       concurrency INTEGER DEFAULT 1,
-      reasoning_effort TEXT
+      reasoning_effort TEXT,
+      mimo_plan TEXT
     );
 
     CREATE TABLE IF NOT EXISTS webdav_configs (
@@ -109,6 +110,7 @@ export function initializeSchema(db: Database.Database): void {
   `);
 
   addColumnIfMissing(db, 'ai_configs', 'reasoning_effort', 'TEXT');
+  addColumnIfMissing(db, 'ai_configs', 'mimo_plan', 'TEXT');
   addColumnIfMissing(db, 'repositories', 'category_locked', 'INTEGER DEFAULT 0');
   addColumnIfMissing(db, 'releases', 'zipball_url', 'TEXT');
   addColumnIfMissing(db, 'releases', 'tarball_url', 'TEXT');
diff --git a/server/src/routes/configs.ts b/server/src/routes/configs.ts
index 15271730..0305dd1c 100644
--- a/server/src/routes/configs.ts
+++ b/server/src/routes/configs.ts
@@ -67,6 +67,7 @@ router.get('/api/configs/ai', (req, res) => {
         useCustomPrompt: !!row.use_custom_prompt,
         concurrency: row.concurrency ?? 1,
         reasoningEffort: row.reasoning_effort ?? null,
+        mimoPlan: row.mimo_plan ?? null,
       };
     });
     res.json(configs);
@@ -80,18 +81,18 @@ router.get('/api/configs/ai', (req, res) => {
 router.post('/api/configs/ai', (req, res) => {
   try {
     const db = getDb();
-    const { name, apiType, model, baseUrl, apiKey, isActive, customPrompt, useCustomPrompt, concurrency, reasoningEffort } = req.body as Record<string, unknown>;
+    const { name, apiType, model, baseUrl, apiKey, isActive, customPrompt, useCustomPrompt, concurrency, reasoningEffort, mimoPlan } = req.body as Record<string, unknown>;
 
     const encryptedKey = apiKey && typeof apiKey === 'string' ? encrypt(apiKey, config.encryptionKey) : null;
 
     const result = db.prepare(
-      'INSERT INTO ai_configs (name, api_type, model, base_url, api_key_encrypted, is_active, custom_prompt, use_custom_prompt, concurrency, reasoning_effort) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?)'
+      'INSERT INTO ai_configs (name, api_type, model, base_url, api_key_encrypted, is_active, custom_prompt, use_custom_prompt, concurrency, reasoning_effort, mimo_plan) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)'
     ).run(
       name ?? '', apiType ?? 'openai', model ?? '', baseUrl ?? null,
-      encryptedKey, isActive ? 1 : 0, customPrompt ?? null, useCustomPrompt ? 1 : 0, concurrency ?? 1, reasoningEffort ?? null
+      encryptedKey, isActive ? 1 : 0, customPrompt ?? null, useCustomPrompt ? 1 : 0, concurrency ?? 1, reasoningEffort ?? null, mimoPlan ?? null
     );
 
-    res.status(201).json({ id: result.lastInsertRowid, name, apiType, model, baseUrl, apiKey: maskApiKey(apiKey as string), isActive: !!isActive, reasoningEffort: reasoningEffort ?? null });
+    res.status(201).json({ id: result.lastInsertRowid, name, apiType, model, baseUrl, apiKey: maskApiKey(apiKey as string), isActive: !!isActive, reasoningEffort: reasoningEffort ?? null, mimoPlan: mimoPlan ?? null });
   } catch (err) {
     logger.errorFromError('configs.createAI', 'POST /api/configs/ai error', err);
     res.status(500).json({ error: 'Failed to create AI config', code: 'CREATE_AI_CONFIG_FAILED' });
@@ -118,6 +119,7 @@ router.put('/api/configs/ai/bulk', (req, res) => {
       useCustomPrompt?: boolean;
       concurrency?: number;
       reasoningEffort?: string;
+      mimoPlan?: string;
     }>;
 
     if (!Array.isArray(configs)) {
@@ -135,8 +137,8 @@ router.put('/api/configs/ai/bulk', (req, res) => {
       db.prepare('DELETE FROM ai_configs').run();
 
       const stmt = db.prepare(`
-        INSERT INTO ai_configs (id, name, api_type, base_url, api_key_encrypted, model, is_active, custom_prompt, use_custom_prompt, concurrency, reasoning_effort)
-        VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
+        INSERT INTO ai_configs (id, name, api_type, base_url, api_key_encrypted, model, is_active, custom_prompt, use_custom_prompt, concurrency, reasoning_effort, mimo_plan)
+        VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
       `);
 
       for (const c of configs) {
@@ -170,7 +172,7 @@ router.put('/api/configs/ai/bulk', (req, res) => {
         stmt.run(
           c.id, c.name ?? '', c.apiType ?? 'openai', c.baseUrl ?? '',
           encryptedKey, c.model ?? '', c.isActive ? 1 : 0,
-          c.customPrompt ?? null, c.useCustomPrompt ? 1 : 0, c.concurrency ?? 1, c.reasoningEffort ?? null
+          c.customPrompt ?? null, c.useCustomPrompt ? 1 : 0, c.concurrency ?? 1, c.reasoningEffort ?? null, c.mimoPlan ?? null
         );
         syncResult.inserted++;
       }
@@ -209,7 +211,7 @@ router.put('/api/configs/ai/:id', (req, res) => {
   try {
     const db = getDb();
     const id = req.params.id;
-    const { name, apiType, model, baseUrl, apiKey, isActive, customPrompt, useCustomPrompt, concurrency, reasoningEffort } = req.body as Record<string, unknown>;
+    const { name, apiType, model, baseUrl, apiKey, isActive, customPrompt, useCustomPrompt, concurrency, reasoningEffort, mimoPlan } = req.body as Record<string, unknown>;
 
     let encryptedKey: string | null = null;
     if (apiKey && typeof apiKey === 'string' && !apiKey.startsWith('***')) {
@@ -221,8 +223,8 @@ router.put('/api/configs/ai/:id', (req, res) => {
     }
 
     const result = db.prepare(
-      'UPDATE ai_configs SET name = ?, api_type = ?, model = ?, base_url = ?, api_key_encrypted = ?, is_active = ?, custom_prompt = ?, use_custom_prompt = ?, concurrency = ?, reasoning_effort = ? WHERE id = ?'
-    ).run(name ?? '', apiType ?? 'openai', model ?? '', baseUrl ?? null, encryptedKey, isActive ? 1 : 0, customPrompt ?? null, useCustomPrompt ? 1 : 0, concurrency ?? 1, reasoningEffort ?? null, id);
+      'UPDATE ai_configs SET name = ?, api_type = ?, model = ?, base_url = ?, api_key_encrypted = ?, is_active = ?, custom_prompt = ?, use_custom_prompt = ?, concurrency = ?, reasoning_effort = ?, mimo_plan = ? WHERE id = ?'
+    ).run(name ?? '', apiType ?? 'openai', model ?? '', baseUrl ?? null, encryptedKey, isActive ? 1 : 0, customPrompt ?? null, useCustomPrompt ? 1 : 0, concurrency ?? 1, reasoningEffort ?? null, mimoPlan ?? null, id);
 
     if (result.changes === 0) {
       res.status(404).json({ error: 'AI config not found', code: 'AI_CONFIG_NOT_FOUND' });
@@ -233,7 +235,7 @@ router.put('/api/configs/ai/:id', (req, res) => {
       try { maskedKey = maskApiKey(decrypt(encryptedKey, config.encryptionKey)); } catch { maskedKey = '****'; }
     }
 
-    res.json({ id, name, apiType, model, baseUrl, apiKey: maskedKey, isActive: !!isActive, reasoningEffort: reasoningEffort ?? null });
+    res.json({ id, name, apiType, model, baseUrl, apiKey: maskedKey, isActive: !!isActive, reasoningEffort: reasoningEffort ?? null, mimoPlan: mimoPlan ?? null });
   } catch (err) {
     logger.errorFromError('configs.updateAI', 'PUT /api/configs/ai error', err);
     res.status(500).json({ error: 'Failed to update AI config', code: 'UPDATE_AI_CONFIG_FAILED' });
diff --git a/server/src/routes/proxy.ts b/server/src/routes/proxy.ts
index bc15a363..217c9bd0 100644
--- a/server/src/routes/proxy.ts
+++ b/server/src/routes/proxy.ts
@@ -176,7 +176,7 @@ router.post('/api/proxy/ai', async (req, res) => {
       'Accept': 'application/json',
     };
 
-    if (apiType === 'openai' || apiType === 'openai-responses' || apiType === 'openai-compatible') {
+    if (apiType === 'openai' || apiType === 'openai-responses' || apiType === 'openai-compatible' || apiType === 'deepseek' || apiType === 'mimo') {
       // openai-compatible 类型直接使用 baseUrl 作为完整地址
       targetUrl = apiType === 'openai-compatible'
         ? baseUrl.replace(/\/$/, '')
@@ -197,11 +197,14 @@ router.post('/api/proxy/ai', async (req, res) => {
       targetUrl = urlObj.toString();
     }
 
+    // DeepSeek Reasoner does not support the reasoning parameter
+    const isDeepSeekReasoner = model.trim() === 'deepseek-reasoner';
     const effectiveRequestBody = (
       reasoningEffort
+      && !isDeepSeekReasoner
       && typeof requestBody === 'object'
       && requestBody !== null
-      && (apiType === 'openai' || apiType === 'openai-responses' || apiType === 'openai-compatible')
+      && (apiType === 'openai' || apiType === 'openai-responses' || apiType === 'openai-compatible' || apiType === 'deepseek' || apiType === 'mimo')
       && !('reasoning' in requestBody)
     )
       ? { ...requestBody, reasoning: { effort: reasoningEffort } }
diff --git a/server/src/routes/sync.ts b/server/src/routes/sync.ts
index 2368449c..b53daf7e 100644
--- a/server/src/routes/sync.ts
+++ b/server/src/routes/sync.ts
@@ -200,13 +200,14 @@ router.post('/api/sync/import', (req, res) => {
           const existingKey = (existing?.api_key_encrypted as string) ?? null;
           // Skip masked keys, keep existing encrypted value
           db.prepare(`
-            INSERT OR REPLACE INTO ai_configs (id, name, api_type, base_url, api_key_encrypted, model, is_active, custom_prompt, use_custom_prompt, concurrency, reasoning_effort)
-            VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
+            INSERT OR REPLACE INTO ai_configs (id, name, api_type, base_url, api_key_encrypted, model, is_active, custom_prompt, use_custom_prompt, concurrency, reasoning_effort, mimo_plan)
+            VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
           `).run(
             c.id, c.name ?? '', c.api_type ?? c.apiType ?? 'openai', c.base_url ?? c.baseUrl ?? null,
             existingKey, c.model ?? '',
             (c.is_active ?? c.isActive) ? 1 : 0, c.custom_prompt ?? c.customPrompt ?? null,
-            (c.use_custom_prompt ?? c.useCustomPrompt) ? 1 : 0, c.concurrency ?? 1, c.reasoning_effort ?? c.reasoningEffort ?? null
+            (c.use_custom_prompt ?? c.useCustomPrompt) ? 1 : 0, c.concurrency ?? 1, c.reasoning_effort ?? c.reasoningEffort ?? null,
+            c.mimo_plan ?? c.mimoPlan ?? null
           );
         }
         counts.ai_configs = aiConfigs.length;
diff --git a/server/src/types/api.ts b/server/src/types/api.ts
index 04741672..5c2b3034 100644
--- a/server/src/types/api.ts
+++ b/server/src/types/api.ts
@@ -67,6 +67,7 @@ export interface AIConfigRow {
   use_custom_prompt: number;
   concurrency: number;
   reasoning_effort: string | null;
+  mimo_plan: string | null;
 }
 
 export interface WebDAVConfigRow {
@@ -128,6 +129,7 @@ export interface SyncAIConfigsRequest {
     useCustomPrompt?: boolean;
     concurrency?: number;
     reasoningEffort?: string;
+    mimoPlan?: string;
   }>;
 }
 
diff --git a/src/components/settings/AIConfigPanel.tsx b/src/components/settings/AIConfigPanel.tsx
index 3a32559b..a1e2575b 100644
--- a/src/components/settings/AIConfigPanel.tsx
+++ b/src/components/settings/AIConfigPanel.tsx
@@ -1,6 +1,6 @@
 import React, { useState, useCallback, useMemo, useRef, useEffect } from 'react';
 import { Bot, Plus, Edit3, Trash2, Save, X, TestTube, RefreshCw, MessageSquare, Eye, EyeOff, AlertCircle } from 'lucide-react';
-import { AIConfig, AIApiType, AIReasoningEffort } from '../../types';
+import { AIConfig, AIApiType, AIReasoningEffort, MiMoPlan } from '../../types';
 import { useAppStore } from '../../store/useAppStore';
 import { AIService } from '../../services/aiService';
 import { buildFinalApiUrl } from '../../utils/apiUrlBuilder';
@@ -21,6 +21,12 @@ type AIFormState = {
   useCustomPrompt: boolean;
   concurrency: number;
   reasoningEffort: '' | AIReasoningEffort;
+  mimoPlan: MiMoPlan;
+};
+
+const MIMO_PLAN_ENDPOINTS: Record<MiMoPlan, string> = {
+  api: 'https://api.xiaomimimo.com/v1',
+  'token-plan': 'https://token-plan-cn.xiaomimimo.com/v1',
 };
 
 const DEFAULT_API_ENDPOINTS: Record<AIApiType, string> = {
@@ -28,9 +34,44 @@ const DEFAULT_API_ENDPOINTS: Record<AIApiType, string> = {
   'openai-responses': 'https://api.openai.com/v1',
   claude: 'https://api.anthropic.com/v1',
   gemini: 'https://generativelanguage.googleapis.com/v1beta',
+  deepseek: 'https://api.deepseek.com',
+  mimo: MIMO_PLAN_ENDPOINTS.api,
   'openai-compatible': '',
 };
 
+function getEndpointPlaceholder(apiType: AIApiType, mimoPlan: MiMoPlan): string {
+  switch (apiType) {
+    case 'openai':
+    case 'openai-responses':
+      return 'https://api.openai.com/v1';
+    case 'claude':
+      return 'https://api.anthropic.com/v1';
+    case 'deepseek':
+      return 'https://api.deepseek.com';
+    case 'mimo':
+      return MIMO_PLAN_ENDPOINTS[mimoPlan];
+    case 'openai-compatible':
+      return 'https://integrate.api.nvidia.com/v1/chat/completions';
+    default:
+      return 'https://generativelanguage.googleapis.com/v1beta';
+  }
+}
+
+function getEndpointHelpText(apiType: AIApiType, t: (zh: string, en: string) => string): string {
+  switch (apiType) {
+    case 'openai-compatible':
+      return t('填写完整的API调用地址，包含完整路径', 'Enter the full API endpoint URL including the complete path');
+    case 'gemini':
+      return t('只填到 v1beta 即可，路径会自动生成', 'Only include the version prefix v1beta, the path will be generated automatically');
+    case 'deepseek':
+      return t('填写到域名即可（如 https://api.deepseek.com），路径会自动生成', 'Only include the domain (e.g. https://api.deepseek.com), the path will be generated automatically');
+    case 'mimo':
+      return t('填写到 /v1 即可（如 https://api.xiaomimimo.com/v1），路径会自动生成', 'Only include up to /v1 (e.g. https://api.xiaomimimo.com/v1), the path will be generated automatically');
+    default:
+      return t('只填到版本号即可（如 .../v1 或 .../v1beta），不要包含 /chat/completions、/responses、/messages', 'Only include the version prefix (e.g. .../v1 or .../v1beta). Do not include /chat/completions, /responses, or /messages.');
+  }
+}
+
 export const AIConfigPanel: React.FC<AIConfigPanelProps> = ({ t }) => {
   const {
     aiConfigs,
@@ -79,6 +120,7 @@ export const AIConfigPanel: React.FC<AIConfigPanelProps> = ({ t }) => {
     useCustomPrompt: false,
     concurrency: 1,
     reasoningEffort: '',
+    mimoPlan: 'api',
   });
 
   // Auto-fill baseUrl when API type changes
@@ -99,6 +141,18 @@ export const AIConfigPanel: React.FC<AIConfigPanelProps> = ({ t }) => {
     }
   }, [form.apiType, form.baseUrl]);
 
+  // Auto-fill baseUrl when MiMo plan changes
+  const prevMimoPlanRef = useRef<MiMoPlan>('api');
+  useEffect(() => {
+    if (form.apiType === 'mimo' && form.mimoPlan !== prevMimoPlanRef.current) {
+      const prevEndpoint = MIMO_PLAN_ENDPOINTS[prevMimoPlanRef.current];
+      if (form.baseUrl === '' || form.baseUrl === prevEndpoint) {
+        setForm(prev => ({ ...prev, baseUrl: MIMO_PLAN_ENDPOINTS[form.mimoPlan] }));
+      }
+      prevMimoPlanRef.current = form.mimoPlan;
+    }
+  }, [form.apiType, form.mimoPlan]);
+
   const resetForm = () => {
     setForm({
       name: '',
@@ -110,12 +164,14 @@ export const AIConfigPanel: React.FC<AIConfigPanelProps> = ({ t }) => {
       useCustomPrompt: false,
       concurrency: 1,
       reasoningEffort: '',
+      mimoPlan: 'api',
     });
     setShowForm(false);
     setEditingId(null);
     setShowCustomPrompt(false);
     setShowDefaultPrompt(false);
     prevApiTypeRef.current = 'openai';
+    prevMimoPlanRef.current = 'api';
   };
 
   const handleSave = () => {
@@ -137,6 +193,7 @@ export const AIConfigPanel: React.FC<AIConfigPanelProps> = ({ t }) => {
           useCustomPrompt: form.useCustomPrompt,
           concurrency: form.concurrency,
           reasoningEffort: form.reasoningEffort || undefined,
+          mimoPlan: form.apiType === 'mimo' ? form.mimoPlan : undefined,
           isActive: existingConfig.isActive,
         };
         updateAIConfig(editingId, updates);
@@ -154,6 +211,7 @@ export const AIConfigPanel: React.FC<AIConfigPanelProps> = ({ t }) => {
         useCustomPrompt: form.useCustomPrompt,
         concurrency: form.concurrency,
         reasoningEffort: form.reasoningEffort || undefined,
+        mimoPlan: form.apiType === 'mimo' ? form.mimoPlan : undefined,
       };
       addAIConfig(config);
     }
@@ -164,6 +222,7 @@ export const AIConfigPanel: React.FC<AIConfigPanelProps> = ({ t }) => {
   const handleEdit = (config: AIConfig) => {
     // Sync ref to prevent auto-fill effect from overwriting loaded config
     prevApiTypeRef.current = config.apiType || 'openai';
+    prevMimoPlanRef.current = config.mimoPlan || 'api';
     setForm({
       name: config.name,
       apiType: config.apiType || 'openai',
@@ -174,6 +233,7 @@ export const AIConfigPanel: React.FC<AIConfigPanelProps> = ({ t }) => {
       useCustomPrompt: config.useCustomPrompt || false,
       concurrency: config.concurrency || 1,
       reasoningEffort: config.reasoningEffort || '',
+      mimoPlan: config.mimoPlan || 'api',
     });
     setEditingId(config.id);
     setShowForm(true);
@@ -392,9 +452,32 @@ Repository information:
                 <option value="openai-responses">OpenAI (Responses)</option>
                 <option value="claude">Claude</option>
                 <option value="gemini">Gemini</option>
+                <option value="deepseek">DeepSeek</option>
+                <option value="mimo">Xiaomi MiMo</option>
                 <option value="openai-compatible">OpenAI Compatible (Custom Endpoint)</option>
               </select>
             </div>
+
+            {form.apiType === 'mimo' && (
+              <div>
+                <label className="block text-sm font-medium text-gray-900 dark:text-text-secondary mb-1">
+                  {t('MiMo 渠道', 'MiMo Channel')} *
+                </label>
+                <select
+                  value={form.mimoPlan}
+                  onChange={(e) => setForm(prev => ({ ...prev, mimoPlan: e.target.value as MiMoPlan }))}
+                  className="w-full px-3 py-2 border border-black/[0.06] dark:border-white/[0.04] rounded-lg bg-white dark:bg-panel-dark text-gray-900 dark:text-text-primary focus:ring-2 focus:ring-brand-violet focus:border-transparent focus:outline-none"
+                >
+                  <option value="api">{t('API（按量付费）', 'API (Pay-as-you-go)')}</option>
+                  <option value="token-plan">{t('Token Plan（订阅制）', 'Token Plan (Subscription)')}</option>
+                </select>
+                <p className="text-xs text-gray-500 dark:text-text-tertiary mt-1">
+                  {form.mimoPlan === 'api'
+                    ? t('API Key 以 sk- 开头，端点 api.xiaomimimo.com', 'API Key starts with sk-, endpoint api.xiaomimimo.com')
+                    : t('API Key 以 tp- 开头，端点 token-plan-cn.xiaomimimo.com', 'API Key starts with tp-, endpoint token-plan-cn.xiaomimimo.com')}
+                </p>
+              </div>
+            )}
             
             <div>
               <label className="block text-sm font-medium text-gray-900 dark:text-text-secondary mb-1">
@@ -405,31 +488,10 @@ Repository information:
                 value={form.baseUrl}
                 onChange={(e) => setForm(prev => ({ ...prev, baseUrl: e.target.value }))}
                 className="w-full px-3 py-2 border border-black/[0.06] dark:border-white/[0.04] rounded-lg bg-white dark:bg-panel-dark text-gray-900 dark:text-text-primary focus:ring-2 focus:ring-brand-violet focus:border-transparent focus:outline-none"
-                placeholder={
-                  form.apiType === 'openai' || form.apiType === 'openai-responses'
-                    ? 'https://api.openai.com/v1'
-                    : form.apiType === 'claude'
-                      ? 'https://api.anthropic.com/v1'
-                      : form.apiType === 'openai-compatible'
-                        ? 'https://integrate.api.nvidia.com/v1/chat/completions'
-                        : 'https://generativelanguage.googleapis.com/v1beta'
-                }
+                placeholder={getEndpointPlaceholder(form.apiType, form.mimoPlan)}
               />
               <p className="text-xs text-gray-500 dark:text-text-tertiary mt-1">
-                {form.apiType === 'openai-compatible'
-                  ? t(
-                      '填写完整的API调用地址，包含完整路径',
-                      'Enter the full API endpoint URL including the complete path'
-                    )
-                  : form.apiType === 'gemini'
-                    ? t(
-                        '只填到 v1beta 即可，路径会自动生成',
-                        'Only include the version prefix v1beta, the path will be generated automatically'
-                      )
-                    : t(
-                        '只填到版本号即可（如 .../v1 或 .../v1beta），不要包含 /chat/completions、/responses、/messages',
-                        'Only include the version prefix (e.g. .../v1 or .../v1beta). Do not include /chat/completions, /responses, or /messages.'
-                      )}
+                {getEndpointHelpText(form.apiType, t)}
               </p>
               {form.baseUrl && (
                 <p className="text-xs text-gray-500 dark:text-text-tertiary mt-1">
diff --git a/src/services/aiService.ts b/src/services/aiService.ts
index 0823c1ef..818a3b9f 100644
--- a/src/services/aiService.ts
+++ b/src/services/aiService.ts
@@ -119,12 +119,25 @@ export class AIService {
     return effort ? { effort } : undefined;
   }
 
+  private isDeepSeekModel(): boolean {
+    return this.getApiType() === 'deepseek';
+  }
+
   private isDeepSeekReasonerModel(): boolean {
-    return this.getApiType() === 'openai' && this.config.model.trim() === 'deepseek-reasoner';
+    return this.isDeepSeekModel() && this.config.model.trim() === 'deepseek-reasoner';
+  }
+
+  /**
+   * Check if the model is a DeepSeek model with default thinking enabled (e.g. deepseek-v4-pro, deepseek-v4-flash).
+   * These models consume max_tokens for reasoning, leaving 0 tokens for content if max_tokens is too low.
+   * We need to explicitly disable thinking for these models.
+   */
+  private isDeepSeekThinkingModel(): boolean {
+    return this.isDeepSeekModel() && this.config.model.trim() !== 'deepseek-reasoner';
   }
 
   private isMiMoModel(): boolean {
-    return this.config.model.trim().toLowerCase().includes('mimo');
+    return this.getApiType() === 'mimo';
   }
 
   private async extractErrorDetail(response: Response): Promise<string> {
@@ -154,7 +167,7 @@ export class AIService {
     const configId = this.config.id;
     const reasoning = this.getOpenAIReasoningPayload();
 
-    if (apiType === 'openai' || apiType === 'openai-responses' || apiType === 'openai-compatible') {
+    if (apiType === 'openai' || apiType === 'openai-responses' || apiType === 'openai-compatible' || apiType === 'deepseek' || apiType === 'mimo') {
       const messages = [
         ...(options.system.trim()
           ? [{ role: 'system', content: options.system }]
@@ -162,6 +175,7 @@ export class AIService {
         { role: 'user', content: options.user },
       ];
       const isDeepSeekReasoner = this.isDeepSeekReasonerModel();
+      const isDeepSeekThinking = this.isDeepSeekThinkingModel();
       const isMiMoModel = this.isMiMoModel();
 
       const requestBody = apiType === 'openai-responses'
@@ -171,15 +185,15 @@ export class AIService {
             temperature: options.temperature,
             max_output_tokens: options.maxTokens,
             ...(reasoning ? { reasoning } : {}),
-            ...(isMiMoModel ? { thinking: { type: 'disabled' } } : {}),
+            ...(isMiMoModel || isDeepSeekThinking ? { thinking: { type: 'disabled' } } : {}),
           }
         : {
             model: this.config.model,
             messages,
             max_tokens: options.maxTokens,
             ...(!isDeepSeekReasoner ? { temperature: options.temperature } : {}),
-            ...(!isDeepSeekReasoner && reasoning && apiType !== 'openai-compatible' ? { reasoning } : {}),
-            ...(isMiMoModel ? { thinking: { type: 'disabled' } } : {}),
+            ...(!isDeepSeekReasoner && !isDeepSeekThinking && !isMiMoModel && reasoning && apiType !== 'openai-compatible' ? { reasoning } : {}),
+            ...(isMiMoModel || isDeepSeekThinking ? { thinking: { type: 'disabled' } } : {}),
           };
 
       let data: Record<string, unknown>;
@@ -264,11 +278,18 @@ export class AIService {
           return content;
         }
 
+        // Only fall back to reasoning_content for the dedicated deepseek-reasoner model.
+        // Other DeepSeek models (e.g. deepseek-v4-flash, deepseek-v4-pro) may also return
+        // reasoning_content (the thinking chain), but we must not use it as the final answer.
         const reasoningContent = message?.reasoning_content;
-        if (reasoningContent) {
+        if (reasoningContent && isDeepSeekReasoner) {
           this.logAIRequestDebug(startTime, { apiType, model, configId }, { responseLength: reasoningContent.length }, httpDetails);
           return reasoningContent;
         }
+
+        if (!content && reasoningContent) {
+          logger.warn('ai', 'Model returned reasoning_content but empty content', { model, configId });
+        }
       }
 
       this.logAIRequestDebug(startTime, { apiType, model, configId }, { error: 'request failed' }, httpDetails);
@@ -479,7 +500,7 @@ ${options.user}` : options.user;
         system,
         user: prompt,
         temperature: 0.3,
-        maxTokens: 700,
+        maxTokens: 1000,
         signal,
       });
 
@@ -587,8 +608,12 @@ ${repoInfo}
 
   private parseAIResponse(content: string): { summary: string; tags: string[]; platforms: string[] } {
     try {
+      // Strip thinking tags that some models embed in the content field (e.g. <think>...</think>)
+      // Also handle truncated tags (dangling <think> without </think>) from token exhaustion
       const cleaned = content
         .trim()
+        .replace(/<think>[\s\S]*?<\/think>/gi, '')
+        .replace(/<think>[\s\S]*$/gi, '')
         .replace(/^```(?:json)?\s*/i, '')
         .replace(/\s*```$/i, '')
         .trim();
diff --git a/src/types/index.ts b/src/types/index.ts
index 94f9d6f7..a0878d57 100644
--- a/src/types/index.ts
+++ b/src/types/index.ts
@@ -127,8 +127,9 @@ export interface GitHubUser {
   email: string | null;
 }
 
-export type AIApiType = 'openai' | 'openai-responses' | 'claude' | 'gemini' | 'openai-compatible';
+export type AIApiType = 'openai' | 'openai-responses' | 'claude' | 'gemini' | 'deepseek' | 'mimo' | 'openai-compatible';
 export type AIReasoningEffort = 'none' | 'low' | 'medium' | 'high' | 'xhigh';
+export type MiMoPlan = 'api' | 'token-plan';
 
 export type SecretStatus = 'ok' | 'empty' | 'decrypt_failed';
 
@@ -144,6 +145,7 @@ export interface AIConfig {
   useCustomPrompt?: boolean; // 是否使用自定义提示词
   concurrency?: number; // AI分析并发数，默认为1
   reasoningEffort?: AIReasoningEffort; // OpenAI GPT-5/Responses 可选 reasoning 强度
+  mimoPlan?: MiMoPlan; // MiMo 渠道：api（按量付费）或 token-plan（订阅制）
   apiKeyStatus?: SecretStatus;
 }
 
diff --git a/src/utils/apiUrlBuilder.ts b/src/utils/apiUrlBuilder.ts
index 0d059df5..57d605be 100644
--- a/src/utils/apiUrlBuilder.ts
+++ b/src/utils/apiUrlBuilder.ts
@@ -44,6 +44,7 @@ export function buildFinalApiUrl(baseUrl: string, apiType: AIApiType): string {
     return buildApiUrl(baseUrl, 'v1beta/models/{model}:generateContent');
   }
 
+  // deepseek uses the same Chat Completions endpoint as openai
   const pathWithVersion = apiType === 'openai-responses' ? 'v1/responses' : 'v1/chat/completions';
   return buildApiUrl(baseUrl, pathWithVersion);
 }