@lobehub/chat 0.151.1 → 0.151.3

package/CHANGELOG.md

@@ -2,6 +2,56 @@
 
 # Changelog
 
+### [Version 0.151.3](https://github.com/lobehub/lobe-chat/compare/v0.151.2...v0.151.3)
+
+<sup>Released on **2024-04-29**</sup>
+
+#### 💄 Styles
+
+- **misc**: Patching models info.
+
+<br/>
+
+<details>
+<summary><kbd>Improvements and Fixes</kbd></summary>
+
+#### Styles
+
+- **misc**: Patching models info, closes [#2269](https://github.com/lobehub/lobe-chat/issues/2269) [#22802280](https://github.com/lobehub/lobe-chat/issues/22802280) ([03bcb06](https://github.com/lobehub/lobe-chat/commit/03bcb06))
+
+</details>
+
+<div align="right">
+
+[![](https://img.shields.io/badge/-BACK_TO_TOP-151515?style=flat-square)](#readme-top)
+
+</div>
+
+### [Version 0.151.2](https://github.com/lobehub/lobe-chat/compare/v0.151.1...v0.151.2)
+
+<sup>Released on **2024-04-29**</sup>
+
+#### 🐛 Bug Fixes
+
+- **misc**: Fix only inject welcome question in inbox.
+
+<br/>
+
+<details>
+<summary><kbd>Improvements and Fixes</kbd></summary>
+
+#### What's fixed
+
+- **misc**: Fix only inject welcome question in inbox, closes [#2289](https://github.com/lobehub/lobe-chat/issues/2289) ([cc8edd3](https://github.com/lobehub/lobe-chat/commit/cc8edd3))
+
+</details>
+
+<div align="right">
+
+[![](https://img.shields.io/badge/-BACK_TO_TOP-151515?style=flat-square)](#readme-top)
+
+</div>
+
 ### [Version 0.151.1](https://github.com/lobehub/lobe-chat/compare/v0.151.0...v0.151.1)
 
 <sup>Released on **2024-04-29**</sup>

package/next.config.mjs

@@ -23,6 +23,7 @@ const nextConfig = {
       '@lobehub/ui',
       'gpt-tokenizer',
       'chroma-js',
+      'shiki',
     ],
     webVitalsAttribution: ['CLS', 'LCP'],
   },

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@lobehub/chat",
-  "version": "0.151.1",
+  "version": "0.151.3",
   "description": "Lobe Chat - an open-source, high-performance chatbot 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",
@@ -126,7 +126,7 @@
     "lucide-react": "latest",
     "modern-screenshot": "^4.4.39",
     "nanoid": "^5.0.7",
-    "next": "14.1.4",
+    "next": "^14.2.1",
     "next-auth": "5.0.0-beta.15",
     "next-sitemap": "^4.2.3",
     "numeral": "^2.0.6",

package/src/app/chat/(desktop)/features/ChatInput/Footer/SendMore.tsx

@@ -45,7 +45,7 @@ const SendMore = memo(() => {
     hotKey,
     (keyboardEvent, hotkeysEvent) => {
       console.log(keyboardEvent, hotkeysEvent);
-      sendMessage(true);
+      sendMessage({ onlyAddUserMessage: true });
     },
     {
       enableOnFormTags: true,
@@ -94,7 +94,7 @@ const SendMore = memo(() => {
               </Flexbox>
             ),
             onClick: () => {
-              sendMessage(true);
+              sendMessage({ onlyAddUserMessage: true });
             },
           },
         ],

package/src/config/modelProviders/anthropic.ts

@@ -1,6 +1,6 @@
 import { ModelProviderCard } from '@/types/llm';
 
-// ref https://docs.anthropic.com/claude/docs/models-overview
+// ref https://docs.anthropic.com/claude/docs/models-overview#model-comparison
 const Anthropic: ModelProviderCard = {
   chatModels: [
     {

package/src/config/modelProviders/azure.ts

@@ -1,5 +1,6 @@
 import { ModelProviderCard } from '@/types/llm';
 
+// ref https://learn.microsoft.com/en-us/azure/ai-services/openai/concepts/models
 const Azure: ModelProviderCard = {
   chatModels: [
     {

package/src/config/modelProviders/bedrock.ts

@@ -56,14 +56,14 @@ const Bedrock: ModelProviderCard = {
       displayName: 'Llama 2 Chat 13B',
       enabled: true,
       id: 'meta.llama2-13b-chat-v1',
-      tokens: 4000,
+      tokens: 4096,
     },
     {
       description: 'Llama 2 Chat 70B v1，上下文大小为 4k，Llama 2 模型的对话用例优化变体。',
       displayName: 'Llama 2 Chat 70B',
       enabled: true,
       id: 'meta.llama2-70b-chat-v1',
-      tokens: 4000,
+      tokens: 4096,
     },
   ],
   id: 'bedrock',

package/src/config/modelProviders/google.ts

@@ -1,6 +1,7 @@
 import { ModelProviderCard } from '@/types/llm';
 
 // ref https://ai.google.dev/models/gemini
+// api https://ai.google.dev/api/rest/v1beta/models/list
 const Google: ModelProviderCard = {
   chatModels: [
     {
@@ -8,13 +9,14 @@ const Google: ModelProviderCard = {
       displayName: 'PaLM 2 Chat (Legacy)',
       id: 'chat-bison-001',
       maxOutput: 1024,
+      // tokens: 4096 + 1024, // none tokens test
     },
     {
       description: 'A legacy model that understands text and generates text as an output',
       displayName: 'PaLM 2 (Legacy)',
       id: 'text-bison-001',
       maxOutput: 1024,
-      tokens: 9220,
+      tokens: 8196 + 1024,
     },
     {
       description: 'The best model for scaling across a wide range of tasks',
@@ -22,14 +24,14 @@ const Google: ModelProviderCard = {
       enabled: true,
       id: 'gemini-pro',
       maxOutput: 2048,
-      tokens: 32_768,
+      tokens: 30_720 + 2048,
     },
     {
       description: 'The best image understanding model to handle a broad range of applications',
       displayName: 'Gemini 1.0 Pro Vision',
       id: 'gemini-1.0-pro-vision-latest',
       maxOutput: 4096,
-      tokens: 16_384,
+      tokens: 12_288 + 4096,
       vision: true,
     },
     {
@@ -38,7 +40,7 @@ const Google: ModelProviderCard = {
       enabled: true,
       id: 'gemini-pro-vision',
       maxOutput: 4096,
-      tokens: 16_384,
+      tokens: 12_288 + 4096,
       vision: true,
     },
     {
@@ -47,7 +49,7 @@ const Google: ModelProviderCard = {
       displayName: 'Gemini 1.0 Pro 001 (Tuning)',
       id: 'gemini-1.0-pro-001',
       maxOutput: 2048,
-      tokens: 32_768,
+      tokens: 30_720 + 2048,
     },
     {
       description:
@@ -55,7 +57,7 @@ const Google: ModelProviderCard = {
       displayName: 'Gemini 1.0 Pro Latest',
       id: 'gemini-1.0-pro-latest',
       maxOutput: 2048,
-      tokens: 32_768,
+      tokens: 30_720 + 2048,
     },
     {
       description: 'Mid-size multimodal model that supports up to 1 million tokens',
@@ -63,7 +65,7 @@ const Google: ModelProviderCard = {
       enabled: true,
       id: 'gemini-1.5-pro-latest',
       maxOutput: 8192,
-      tokens: 1_056_768,
+      tokens: 1_048_576 + 8192,
       vision: true,
     },
     {

package/src/config/modelProviders/groq.ts

@@ -29,7 +29,6 @@ const Groq: ModelProviderCard = {
     },
     {
       displayName: 'LLaMA2-70b-chat',
-      enabled: true,
       id: 'llama2-70b-4096',
       tokens: 4096,
     },

package/src/config/modelProviders/minimax.ts

@@ -1,7 +1,22 @@
 import { ModelProviderCard } from '@/types/llm';
 
+// ref https://www.minimaxi.com/document/guides/chat-model/pro/api
 const Minimax: ModelProviderCard = {
   chatModels: [
+    {
+      description: '复杂场景，例如应用题计算、科学计算等场景',
+      displayName: 'abab6.5',
+      enabled: true,
+      id: 'abab6.5-chat',
+      tokens: 8192,
+    },
+    {
+      description: '通用场景',
+      displayName: 'abab6.5s',
+      enabled: true,
+      id: 'abab6.5s-chat',
+      tokens: 245_760,
+    },
     {
       description: '更复杂的格式化文本生成',
       displayName: 'abab6',

package/src/config/modelProviders/mistral.ts

@@ -1,6 +1,6 @@
 import { ModelProviderCard } from '@/types/llm';
 
-// ref https://docs.mistral.ai/platform/pricing/#chat-completions-api
+// ref https://docs.mistral.ai/getting-started/models/
 const Mistral: ModelProviderCard = {
   chatModels: [
     {

package/src/config/modelProviders/moonshot.ts

@@ -1,5 +1,6 @@
 import { ModelProviderCard } from '@/types/llm';
 
+// ref https://platform.moonshot.cn/docs/intro#模型列表
 const Moonshot: ModelProviderCard = {
   chatModels: [
     {

package/src/config/modelProviders/ollama.ts

@@ -6,7 +6,7 @@ const Ollama: ModelProviderCard = {
       displayName: 'Llama3 8B',
       enabled: true,
       id: 'llama3',
-      tokens: 8000,
+      tokens: 8000, // https://huggingface.co/blog/zh/llama3#llama-3-的新进展
     },
     {
       displayName: 'Llama3 70B',
@@ -17,48 +17,48 @@ const Ollama: ModelProviderCard = {
       displayName: 'Command R 35B',
       enabled: true,
       id: 'command-r',
-      tokens: 128_000,
+      tokens: 131_072, // https://huggingface.co/CohereForAI/c4ai-command-r-v01/blob/main/config.json
     },
     {
       displayName: 'Command R+ 104B (Q2_K)',
       id: 'command-r-plus:104b-q2_K',
-      tokens: 128_000,
+      tokens: 131_072, // https://huggingface.co/CohereForAI/c4ai-command-r-plus/blob/main/config.json
     },
     {
       displayName: 'Gemma 7B',
       enabled: true,
       id: 'gemma',
-      tokens: 4000,
+      tokens: 8192, // https://huggingface.co/google/gemma-7b-it/discussions/73#65e9678c0cda621164a95bad
     },
     {
       displayName: 'Gemma 2B',
       id: 'gemma:2b',
-      tokens: 4000,
+      tokens: 8192,
     },
     {
       displayName: 'Llama2 Chat 13B',
       id: 'llama2:13b',
-      tokens: 4000,
+      tokens: 4096, // https://llama.meta.com/llama2/
     },
     {
       displayName: 'Llama2 Chat 7B',
       id: 'llama2',
-      tokens: 4000,
+      tokens: 4096,
     },
     {
       displayName: 'Llama2 Chat 70B',
       id: 'llama2:70b',
-      tokens: 4000,
+      tokens: 4096,
     },
     {
       displayName: 'Llama2 CN 13B',
       id: 'llama2-chinese:13b',
-      tokens: 4000,
+      tokens: 4096,
     },
     {
       displayName: 'Llama2 CN 7B',
       id: 'llama2-chinese',
-      tokens: 4000,
+      tokens: 4096,
     },
     {
       displayName: 'WizardLM 2 7B',
@@ -74,45 +74,45 @@ const Ollama: ModelProviderCard = {
     {
       displayName: 'Code Llama 7B',
       id: 'codellama',
-      tokens: 16_000,
+      tokens: 16_384, // https://huggingface.co/codellama/CodeLlama-7b-hf/blob/main/config.json
     },
     {
       displayName: 'Code Llama 34B',
       id: 'codellama:34b',
-      tokens: 16_000,
+      tokens: 16_384,
     },
     {
       displayName: 'Code Llama 70B',
       id: 'codellama:70b',
-      tokens: 16_000,
+      tokens: 16_384,
     },
     {
       displayName: 'Code Llama 7B (Python)',
       id: 'codellama:python',
-      tokens: 16_000,
+      tokens: 16_384,
     },
     {
       displayName: 'Phi3-Instruct 3.8B',
       enabled: true,
       id: 'phi3:instruct',
-      tokens: 128_000,
+      tokens: 131_072, // https://huggingface.co/microsoft/Phi-3-mini-128k-instruct/blob/main/config.json
     },
     {
       displayName: 'Mistral',
       enabled: true,
       id: 'mistral',
-      tokens: 4800,
+      tokens: 32_768, // https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2/blob/main/config.json
     },
     {
       displayName: 'Mixtral 8x7B',
       enabled: true,
       id: 'mixtral',
-      tokens: 32_000,
+      tokens: 32_768,
     },
     {
       displayName: 'Mixtral 8x22B',
       id: 'mixtral:8x22b',
-      tokens: 64_000,
+      tokens: 65_536, // https://huggingface.co/mistralai/Mixtral-8x22B-v0.1/blob/main/config.json
     },
     {
       displayName: 'Qwen Chat 4B',
@@ -138,19 +138,19 @@ const Ollama: ModelProviderCard = {
     {
       displayName: 'LLaVA 7B',
       id: 'llava',
-      tokens: 4000,
+      tokens: 4096, // https://huggingface.co/llava-hf/llava-1.5-7b-hf/blob/main/config.json
       vision: true,
     },
     {
       displayName: 'LLaVA 13B',
       id: 'llava:13b',
-      tokens: 4000,
+      tokens: 4096,
       vision: true,
     },
     {
       displayName: 'LLaVA 34B',
       id: 'llava:34b',
-      tokens: 4000,
+      tokens: 4096,
       vision: true,
     },
   ],

package/src/config/modelProviders/openai.ts

@@ -1,6 +1,6 @@
 import { ModelProviderCard } from '@/types/llm';
 
-// refs to: https://platform.openai.com/docs/models/gpt-4-turbo-and-gpt-4
+// ref https://platform.openai.com/docs/models
 const OpenAI: ModelProviderCard = {
   chatModels: [
     {

package/src/config/modelProviders/togetherai.ts

@@ -1,6 +1,6 @@
 import { ModelProviderCard } from '@/types/llm';
 
-// ref https://api.together.xyz/models
+// ref https://docs.together.ai/docs/inference-models
 const TogetherAI: ModelProviderCard = {
   chatModels: [
     {

package/src/config/modelProviders/zeroone.ts

@@ -1,5 +1,6 @@
 import { ModelProviderCard } from '@/types/llm';
 
+// ref https://platform.lingyiwanwu.com/
 const ZeroOne: ModelProviderCard = {
   chatModels: [
     {
@@ -7,7 +8,7 @@ const ZeroOne: ModelProviderCard = {
       displayName: 'YI 34B Chat',
       enabled: true,
       id: 'yi-34b-chat-0205',
-      tokens: 4000,
+      tokens: 4096, // https://huggingface.co/01-ai/Yi-34B-Chat/blob/main/config.json
     },
     {
       description:
@@ -15,7 +16,7 @@ const ZeroOne: ModelProviderCard = {
       displayName: 'YI Vision Plus',
       enabled: true,
       id: 'yi-vl-plus',
-      tokens: 4000,
+      tokens: 4096,
       vision: true,
     },
     {
@@ -23,7 +24,7 @@ const ZeroOne: ModelProviderCard = {
       displayName: 'YI 34B Chat 200k',
       enabled: true,
       id: 'yi-34b-chat-200k',
-      tokens: 200_000,
+      tokens: 200_000, // https://huggingface.co/01-ai/Yi-34B-200K/blob/main/config.json
     },
   ],
   id: 'zeroone',

package/src/config/modelProviders/zhipu.ts

@@ -1,8 +1,10 @@
 import { ModelProviderCard } from '@/types/llm';
 
 // TODO: 等待 ZhiPu 修复 API 问题后开启 functionCall
-// refs: https://github.com/lobehub/lobe-chat/discussions/737#discussioncomment-8315815
 // 暂时不透出 GLM 系列的 function_call 功能
+// refs https://github.com/lobehub/lobe-chat/discussions/737#discussioncomment-8315815
+
+// ref https://open.bigmodel.cn/dev/howuse/model
 const ZhiPu: ModelProviderCard = {
   chatModels: [
     {
@@ -19,7 +21,7 @@ const ZhiPu: ModelProviderCard = {
       displayName: 'GLM-4 Vision',
       enabled: true,
       id: 'glm-4v',
-      tokens: 128_000,
+      tokens: 2000,
       vision: true,
     },
     {

package/src/features/ChatInput/useSend.ts

@@ -1,15 +1,21 @@
 import { useCallback } from 'react';
 
 import { useChatStore } from '@/store/chat';
+import { SendMessageParams } from '@/store/chat/slices/message/action';
 import { filesSelectors, useFileStore } from '@/store/file';
 
+export type UseSendMessageParams = Pick<
+  SendMessageParams,
+  'onlyAddUserMessage' | 'isWelcomeQuestion'
+>;
+
 export const useSendMessage = () => {
   const [sendMessage, updateInputMessage] = useChatStore((s) => [
     s.sendMessage,
     s.updateInputMessage,
   ]);
 
-  return useCallback((onlyAddUserMessage?: boolean) => {
+  return useCallback((params: UseSendMessageParams = {}) => {
     const store = useChatStore.getState();
     if (!!store.chatLoadingId) return;
     if (!store.inputMessage) return;
@@ -19,7 +25,7 @@ export const useSendMessage = () => {
     sendMessage({
       files: imageList,
       message: store.inputMessage,
-      onlyAddUserMessage: onlyAddUserMessage,
+      ...params,
     });
 
     updateInputMessage('');

package/src/features/Conversation/components/InboxWelcome/QuestionSuggest.tsx

@@ -10,7 +10,8 @@ import { useTranslation } from 'react-i18next';
 import { Flexbox } from 'react-layout-kit';
 
 import { USAGE_DOCUMENTS } from '@/const/url';
-import { useChatInput } from '@/features/ChatInput/useChatInput';
+import { useSendMessage } from '@/features/ChatInput/useSend';
+import { useChatStore } from '@/store/chat';
 
 const useStyles = createStyles(({ css, token }) => ({
   card: css`
@@ -54,14 +55,10 @@ const qa = shuffle([
 ]).slice(0, 5);
 
 const QuestionSuggest = memo(() => {
-  const { onInput, onSend } = useChatInput();
+  const [updateInputMessage] = useChatStore((s) => [s.updateInputMessage]);
   const { t } = useTranslation('welcome');
   const { styles } = useStyles();
-
-  const handoleSend = (qa: string) => {
-    onInput(qa);
-    onSend();
-  };
+  const sendMessage = useSendMessage();
 
   return (
     <Flexbox gap={8} width={'100%'}>
@@ -85,7 +82,10 @@ const QuestionSuggest = memo(() => {
               gap={8}
               horizontal
               key={item}
-              onClick={() => handoleSend(text)}
+              onClick={() => {
+                updateInputMessage(text);
+                sendMessage({ isWelcomeQuestion: true });
+              }}
             >
               {t(text)}
             </Flexbox>

package/src/features/Conversation/components/VirtualizedList/index.tsx

@@ -7,9 +7,11 @@ import { useChatStore } from '@/store/chat';
 import { chatSelectors } from '@/store/chat/selectors';
 import { isMobileScreen } from '@/utils/screen';
 
+import { useInitConversation } from '../../hooks/useInitConversation';
 import AutoScroll from '../AutoScroll';
 import Item from '../ChatItem';
 import InboxWelcome from '../InboxWelcome';
+import SkeletonList from '../SkeletonList';
 
 const WELCOME_ID = 'welcome';
 
@@ -29,6 +31,8 @@ interface VirtualizedListProps {
   mobile?: boolean;
 }
 const VirtualizedList = memo<VirtualizedListProps>(({ mobile }) => {
+  useInitConversation();
+
   const virtuosoRef = useRef<VirtuosoHandle>(null);
   const [atBottom, setAtBottom] = useState(true);
 
@@ -52,7 +56,9 @@ const VirtualizedList = memo<VirtualizedListProps>(({ mobile }) => {
   // overscan should be 1.5 times the height of the window
   const overscan = typeof window !== 'undefined' ? window.innerHeight * 1.5 : 0;
 
-  return chatLoading && data.length === 2 ? null : (
+  return chatLoading ? (
+    <SkeletonList mobile={mobile} />
+  ) : (
     <Flexbox height={'100%'}>
       <Virtuoso
         atBottomStateChange={setAtBottom}

package/src/features/Conversation/index.tsx

@@ -1,13 +1,12 @@
 import { createStyles } from 'antd-style';
-import { ReactNode, memo } from 'react';
+import { ReactNode, Suspense, lazy, memo } from 'react';
 import { Flexbox } from 'react-layout-kit';
 
 import ChatHydration from '@/components/StoreHydration/ChatHydration';
-import { useChatStore } from '@/store/chat';
 
 import SkeletonList from './components/SkeletonList';
-import ChatList from './components/VirtualizedList';
-import { useInitConversation } from './hooks/useInitConversation';
+
+const ChatList = lazy(() => import('./components/VirtualizedList'));
 
 const useStyles = createStyles(
   ({ css, responsive, stylish }) => css`
@@ -30,10 +29,6 @@ interface ConversationProps {
 const Conversation = memo<ConversationProps>(({ chatInput, mobile }) => {
   const { styles } = useStyles();
 
-  useInitConversation();
-
-  const [messagesInit] = useChatStore((s) => [s.messagesInit]);
-
   return (
     <Flexbox
       flex={1}
@@ -41,7 +36,9 @@ const Conversation = memo<ConversationProps>(({ chatInput, mobile }) => {
       style={{ position: 'relative' }}
     >
       <div className={styles}>
-        {messagesInit ? <ChatList mobile={mobile} /> : <SkeletonList mobile={mobile} />}
+        <Suspense fallback={<SkeletonList mobile={mobile} />}>
+          <ChatList mobile={mobile} />
+        </Suspense>
       </div>
       {chatInput}
       <ChatHydration />

package/src/migrations/FromV3ToV4/fixtures/ollama-output-v4.json

@@ -48,7 +48,7 @@
               "displayName": "LLaVA 7B",
               "enabled": true,
               "id": "llava",
-              "tokens": 4000,
+              "tokens": 4096,
               "vision": true
             }
           ]

package/src/services/chat.ts

@@ -31,6 +31,7 @@ import { createHeaderWithAuth, getProviderAuthPayload } from './_auth';
 import { API_ENDPOINTS } from './_url';
 
 interface FetchOptions {
+  isWelcomeQuestion?: boolean;
   signal?: AbortSignal | undefined;
   trace?: TracePayload;
 }
@@ -65,6 +66,7 @@ interface FetchAITaskResultParams {
 
 interface CreateAssistantMessageStream extends FetchSSEOptions {
   abortController?: AbortController;
+  isWelcomeQuestion?: boolean;
   params: GetChatCompletionPayload;
   trace?: TracePayload;
 }
@@ -220,10 +222,12 @@ class ChatService {
     onErrorHandle,
     onFinish,
     trace,
+    isWelcomeQuestion,
   }: CreateAssistantMessageStream) => {
     await fetchSSE(
       () =>
         this.createAssistantMessage(params, {
+          isWelcomeQuestion,
           signal: abortController?.signal,
           trace: this.mapTrace(trace, TraceTagMap.Chat),
         }),
@@ -432,9 +436,11 @@ class ChatService {
     });
 
     return produce(postMessages, (draft) => {
-      // Inject InboxGuide SystemRole
+      // if it's a welcome question, inject InboxGuide SystemRole
       const inboxGuideSystemRole =
-        options?.trace?.sessionId === INBOX_SESSION_ID && INBOX_GUIDE_SYSTEMROLE;
+        options?.isWelcomeQuestion &&
+        options?.trace?.sessionId === INBOX_SESSION_ID &&
+        INBOX_GUIDE_SYSTEMROLE;
 
       // Inject Tool SystemRole
       const hasTools = tools && tools?.length > 0;

package/src/store/chat/slices/message/action.test.ts

@@ -385,11 +385,7 @@ describe('chatMessage actions', () => {
       });
 
       expect(messageService.removeMessage).not.toHaveBeenCalledWith(messageId);
-      expect(mockState.coreProcessMessage).toHaveBeenCalledWith(
-        expect.any(Array),
-        messageId,
-        undefined,
-      );
+      expect(mockState.coreProcessMessage).toHaveBeenCalledWith(expect.any(Array), messageId, {});
     });
 
     it('should not perform any action if the message id does not exist', async () => {

package/src/store/chat/slices/message/action.ts

@@ -29,10 +29,20 @@ const n = setNamespace('message');
 
 const SWR_USE_FETCH_MESSAGES = 'SWR_USE_FETCH_MESSAGES';
 
-interface SendMessageParams {
+export interface SendMessageParams {
   message: string;
   files?: { id: string; url: string }[];
   onlyAddUserMessage?: boolean;
+  /**
+   *
+   * https://github.com/lobehub/lobe-chat/pull/2086
+   */
+  isWelcomeQuestion?: boolean;
+}
+
+interface ProcessMessageParams {
+  traceId?: string;
+  isWelcomeQuestion?: boolean;
 }
 
 export interface ChatMessageAction {
@@ -77,7 +87,7 @@ export interface ChatMessageAction {
   coreProcessMessage: (
     messages: ChatMessage[],
     parentId: string,
-    traceId?: string,
+    params?: ProcessMessageParams,
   ) => Promise<void>;
   /**
    * 实际获取 AI 响应
@@ -87,7 +97,7 @@ export interface ChatMessageAction {
   fetchAIChatMessage: (
     messages: ChatMessage[],
     assistantMessageId: string,
-    traceId?: string,
+    params?: ProcessMessageParams,
   ) => Promise<{
     content: string;
     functionCallAtEnd: boolean;
@@ -173,7 +183,7 @@ export const chatMessage: StateCreator<
     await messageService.removeAllMessages();
     await refreshMessages();
   },
-  sendMessage: async ({ message, files, onlyAddUserMessage }) => {
+  sendMessage: async ({ message, files, onlyAddUserMessage, isWelcomeQuestion }) => {
     const { coreProcessMessage, activeTopicId, activeId } = get();
     if (!activeId) return;
 
@@ -200,7 +210,7 @@ export const chatMessage: StateCreator<
     // Get the current messages to generate AI response
     const messages = chatSelectors.currentChats(get());
 
-    await coreProcessMessage(messages, id);
+    await coreProcessMessage(messages, id, { isWelcomeQuestion });
 
     // check activeTopic and then auto create topic
     const chats = chatSelectors.currentChats(get());
@@ -263,6 +273,8 @@ export const chatMessage: StateCreator<
       async ([, sessionId, topicId]: [string, string, string | undefined]) =>
         messageService.getMessages(sessionId, topicId),
       {
+        suspense: true,
+        fallbackData: [],
         onSuccess: (messages, key) => {
           set(
             { activeId: sessionId, messages, messagesInit: true },
@@ -280,7 +292,7 @@ export const chatMessage: StateCreator<
   },
 
   // the internal process method of the AI message
-  coreProcessMessage: async (messages, userMessageId, trace) => {
+  coreProcessMessage: async (messages, userMessageId, params) => {
     const { fetchAIChatMessage, triggerFunctionCall, refreshMessages, activeTopicId } = get();
 
     const { model, provider } = getAgentConfig();
@@ -301,7 +313,7 @@ export const chatMessage: StateCreator<
 
     // 2. fetch the AI response
     const { isFunctionCall, content, functionCallAtEnd, functionCallContent, traceId } =
-      await fetchAIChatMessage(messages, mid, trace);
+      await fetchAIChatMessage(messages, mid, params);
 
     // 3. if it's the function call message, trigger the function method
     if (isFunctionCall) {
@@ -341,7 +353,7 @@ export const chatMessage: StateCreator<
 
     set({ messages }, false, n(`dispatchMessage/${payload.type}`, payload));
   },
-  fetchAIChatMessage: async (messages, assistantId, traceId) => {
+  fetchAIChatMessage: async (messages, assistantId, params) => {
     const {
       toggleChatLoading,
       refreshMessages,
@@ -421,11 +433,12 @@ export const chatMessage: StateCreator<
         plugins: config.plugins,
       },
       trace: {
-        traceId,
+        traceId: params?.traceId,
         sessionId: get().activeId,
         topicId: get().activeTopicId,
         traceName: TraceNameMap.Conversation,
       },
+      isWelcomeQuestion: params?.isWelcomeQuestion,
       onErrorHandle: async (error) => {
         await messageService.updateMessageError(assistantId, error);
         await refreshMessages();
@@ -567,7 +580,7 @@ export const chatMessage: StateCreator<
 
     if (!latestMsg) return;
 
-    await coreProcessMessage(contextMessages, latestMsg.id, traceId);
+    await coreProcessMessage(contextMessages, latestMsg.id, { traceId });
   },
 
   internalUpdateMessageContent: async (id, content) => {

package/src/store/chat/slices/plugin/action.test.ts

@@ -61,7 +61,7 @@ describe('ChatPluginAction', () => {
       expect(result.current.coreProcessMessage).toHaveBeenCalledWith(
         mockCurrentChats,
         messageId,
-        undefined,
+        {},
       );
     });
     it('should update message content and not trigger ai message', async () => {

package/src/store/chat/slices/plugin/action.ts

@@ -174,7 +174,7 @@ export const chatPlugin: StateCreator<
   triggerAIMessage: async (id, traceId) => {
     const { coreProcessMessage } = get();
     const chats = chatSelectors.currentChats(get());
-    await coreProcessMessage(chats, id, traceId);
+    await coreProcessMessage(chats, id, { traceId });
   },
 
   triggerFunctionCall: async (id) => {

package/src/store/user/slices/settings/actions/llm.test.ts

@@ -109,7 +109,7 @@ describe('LLMSettingsSliceAction', () => {
         displayName: 'LLaVA 7B',
         enabled: true,
         id: 'llava',
-        tokens: 4000,
+        tokens: 4096,
         vision: true,
       });
     });

package/src/types/llm.ts

@@ -32,7 +32,7 @@ export interface ChatModelCard {
   legacy?: boolean;
   maxOutput?: number;
   /**
-   * the context window
+   * the context window (or input + output tokens limit)
    */
   tokens?: number;
   /**