@lobehub/lobehub 2.0.0-next.136 → 2.0.0-next.137

package/CHANGELOG.md

@@ -2,6 +2,31 @@
 
 # Changelog
 
+## [Version 2.0.0-next.137](https://github.com/lobehub/lobe-chat/compare/v2.0.0-next.136...v2.0.0-next.137)
+
+<sup>Released on **2025-11-30**</sup>
+
+#### 🐛 Bug Fixes
+
+- **misc**: Update apiMode handling in ChatService to prioritize user preferences.
+
+<br/>
+
+<details>
+<summary><kbd>Improvements and Fixes</kbd></summary>
+
+#### What's fixed
+
+- **misc**: Update apiMode handling in ChatService to prioritize user preferences, closes [#10487](https://github.com/lobehub/lobe-chat/issues/10487) ([5483d91](https://github.com/lobehub/lobe-chat/commit/5483d91))
+
+</details>
+
+<div align="right">
+
+[![](https://img.shields.io/badge/-BACK_TO_TOP-151515?style=flat-square)](#readme-top)
+
+</div>
+
 ## [Version 2.0.0-next.136](https://github.com/lobehub/lobe-chat/compare/v2.0.0-next.135...v2.0.0-next.136)
 
 <sup>Released on **2025-11-30**</sup>

package/changelog/v1.json

@@ -1,4 +1,13 @@
 [
+  {
+    "children": {
+      "fixes": [
+        "Update apiMode handling in ChatService to prioritize user preferences."
+      ]
+    },
+    "date": "2025-11-30",
+    "version": "2.0.0-next.137"
+  },
   {
     "children": {
       "fixes": [

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@lobehub/lobehub",
-  "version": "2.0.0-next.136",
+  "version": "2.0.0-next.137",
   "description": "LobeHub - an open-source,comprehensive AI Agent framework that supports speech synthesis, multimodal, and extensible Function Call plugin system. Supports one-click free deployment of your private ChatGPT/LLM web application.",
   "keywords": [
     "framework",

package/packages/model-runtime/src/core/openaiCompatibleFactory/index.ts

@@ -235,18 +235,45 @@ export const createOpenAICompatibleRuntime = <T extends Record<string, any> = an
 
       const log = debug(`${this.logPrefix}:shouldUseResponsesAPI`);
 
-      // Priority 1: User explicitly set apiMode via switch
-      if (userApiMode === 'responses') {
-        log('using Responses API: explicit userApiMode=%s', userApiMode);
+      // Priority 0: Check built-in responsesAPIModels FIRST (highest priority)
+      // These models MUST use Responses API regardless of user settings
+      if (model && responsesAPIModels.has(model)) {
+        log('using Responses API: model %s in built-in responsesAPIModels (forced)', model);
         return true;
       }
 
-      // Priority 2: userApiMode is explicitly set to something else
-      if (userApiMode !== undefined) {
+      // Priority 1: userApiMode is explicitly set to 'chatCompletion' (user disabled the switch)
+      if (userApiMode === 'chatCompletion') {
         log('using Chat Completions API: userApiMode=%s', userApiMode);
         return false;
       }
 
+      // Priority 2: When user enables the switch (userApiMode === 'responses')
+      // Check if useResponseModels is configured - if so, only matching models use Responses API
+      // If useResponseModels is not configured, all models use Responses API
+      if (userApiMode === 'responses') {
+        if (model && flagUseResponseModels?.length) {
+          const matches = flagUseResponseModels.some((m: string | RegExp) =>
+            typeof m === 'string' ? model.includes(m) : (m as RegExp).test(model),
+          );
+          if (matches) {
+            log(
+              'using Responses API: userApiMode=responses and model %s matches useResponseModels',
+              model,
+            );
+            return true;
+          }
+          log(
+            'using Chat Completions API: userApiMode=responses but model %s does not match useResponseModels',
+            model,
+          );
+          return false;
+        }
+        // No useResponseModels configured, use Responses API for all models
+        log('using Responses API: userApiMode=responses (no useResponseModels filter)');
+        return true;
+      }
+
       // Priority 3: Explicit responseApi flag
       if (responseApi) {
         log('using Responses API: explicit responseApi flag for %s', context);
@@ -259,7 +286,7 @@ export const createOpenAICompatibleRuntime = <T extends Record<string, any> = an
         return true;
       }
 
-      // Priority 5: Check if model matches useResponseModels patterns
+      // Priority 5: Check if model matches useResponseModels patterns (without user switch)
       if (model && flagUseResponseModels?.length) {
         const matches = flagUseResponseModels.some((m: string | RegExp) =>
           typeof m === 'string' ? model.includes(m) : (m as RegExp).test(model),
@@ -270,12 +297,6 @@ export const createOpenAICompatibleRuntime = <T extends Record<string, any> = an
         }
       }
 
-      // Priority 6: Check built-in responsesAPIModels
-      if (model && responsesAPIModels.has(model)) {
-        log('using Responses API: model %s in built-in responsesAPIModels', model);
-        return true;
-      }
-
       log('using Chat Completions API for %s', context);
       return false;
     }

package/src/services/chat/chat.test.ts

@@ -1062,7 +1062,7 @@ describe('ChatService', () => {
       );
     });
 
-    it('should make a POST request without response in non-openai provider payload', async () => {
+    it('should make a POST request with chatCompletion apiMode in non-openai provider payload', async () => {
       const params: Partial<ChatStreamPayload> = {
         model: 'deepseek-reasoner',
         provider: 'deepseek',
@@ -1076,6 +1076,7 @@ describe('ChatService', () => {
         stream: true,
         ...DEFAULT_AGENT_CONFIG.params,
         messages: [],
+        apiMode: 'chatCompletion',
         provider: undefined,
       };
 

package/src/services/chat/index.ts

@@ -267,11 +267,14 @@ class ChatService {
       model = findDeploymentName(model, provider);
     }
 
-    const apiMode = aiProviderSelectors.isProviderEnableResponseApi(provider)(
-      getAiInfraStoreState(),
-    )
+    // When user explicitly disables Responses API, set apiMode to 'chatCompletion'
+    // This ensures the user's preference takes priority over provider's useResponseModels config
+    // When user enables Responses API, set to 'responses' to force use Responses API
+    const apiMode: 'responses' | 'chatCompletion' = aiProviderSelectors.isProviderEnableResponseApi(
+      provider,
+    )(getAiInfraStoreState())
       ? 'responses'
-      : undefined;
+      : 'chatCompletion';
 
     // Get the chat config to check streaming preference
     const chatConfig = agentChatConfigSelectors.currentChatConfig(getAgentStoreState());