@lobehub/chat 1.51.2 → 1.51.3

package/CHANGELOG.md

@@ -2,6 +2,47 @@
 
 # Changelog
 
+### [Version 1.51.3](https://github.com/lobehub/lobe-chat/compare/v1.51.2...v1.51.3)
+
+<sup>Released on **2025-02-05**</sup>
+
+#### ♻ Code Refactoring
+
+- **misc**: Refactor Wenxin with LobeOpenAICompatibleFactory.
+
+#### 🐛 Bug Fixes
+
+- **misc**: Artifact Parsing and Rendering Bug Fix for Gemini 2.0 Flash.
+
+#### 💄 Styles
+
+- **misc**: Add Cache, Metadata, FeatureFlag Viewer to DevPanel.
+
+<br/>
+
+<details>
+<summary><kbd>Improvements and Fixes</kbd></summary>
+
+#### Code refactoring
+
+- **misc**: Refactor Wenxin with LobeOpenAICompatibleFactory, closes [#5729](https://github.com/lobehub/lobe-chat/issues/5729) ([a90a75e](https://github.com/lobehub/lobe-chat/commit/a90a75e))
+
+#### What's fixed
+
+- **misc**: Artifact Parsing and Rendering Bug Fix for Gemini 2.0 Flash, closes [#5633](https://github.com/lobehub/lobe-chat/issues/5633) ([7d782b1](https://github.com/lobehub/lobe-chat/commit/7d782b1))
+
+#### Styles
+
+- **misc**: Add Cache, Metadata, FeatureFlag Viewer to DevPanel, closes [#5764](https://github.com/lobehub/lobe-chat/issues/5764) ([db4e9c7](https://github.com/lobehub/lobe-chat/commit/db4e9c7))
+
+</details>
+
+<div align="right">
+
+[![](https://img.shields.io/badge/-BACK_TO_TOP-151515?style=flat-square)](#readme-top)
+
+</div>
+
 ### [Version 1.51.2](https://github.com/lobehub/lobe-chat/compare/v1.51.1...v1.51.2)
 
 <sup>Released on **2025-02-05**</sup>

package/Dockerfile

@@ -212,7 +212,7 @@ ENV \
     # Upstage
     UPSTAGE_API_KEY="" UPSTAGE_MODEL_LIST="" \
     # Wenxin
-    WENXIN_ACCESS_KEY="" WENXIN_SECRET_KEY="" WENXIN_MODEL_LIST="" \
+    WENXIN_API_KEY="" WENXIN_MODEL_LIST="" \
     # xAI
     XAI_API_KEY="" XAI_MODEL_LIST="" XAI_PROXY_URL="" \
     # 01.AI

package/Dockerfile.database

@@ -247,7 +247,7 @@ ENV \
     # Upstage
     UPSTAGE_API_KEY="" UPSTAGE_MODEL_LIST="" \
     # Wenxin
-    WENXIN_ACCESS_KEY="" WENXIN_SECRET_KEY="" WENXIN_MODEL_LIST="" \
+    WENXIN_API_KEY="" WENXIN_MODEL_LIST="" \
     # xAI
     XAI_API_KEY="" XAI_MODEL_LIST="" XAI_PROXY_URL="" \
     # 01.AI

package/changelog/v1.json

@@ -1,4 +1,16 @@
 [
+  {
+    "children": {
+      "improvements": [
+        "Add Cache, Metadata, FeatureFlag Viewer to DevPanel."
+      ],
+      "fixes": [
+        "Artifact Parsing and Rendering Bug Fix for Gemini 2.0 Flash."
+      ]
+    },
+    "date": "2025-02-05",
+    "version": "1.51.3"
+  },
   {
     "children": {
       "improvements": [

package/docs/usage/providers/wenxin.mdx

@@ -22,27 +22,30 @@ This article will guide you on how to use Wenxin Qianfan in LobeChat.
 <Steps>
   ### Step 1: Obtain the Wenxin Qianfan API Key
 
-  - Register and log in to the [Baidu Intelligent Cloud Console](https://console.bce.baidu.com/)
-  - Navigate to `Baidu Intelligent Cloud Qianfan ModelBuilder`
-  - Choose `Application Access` from the left-side menu
-  - Create an application
+  - Register and log in to the [Baidu AI Cloud Console](https://console.bce.baidu.com/)
+  - Navigate to `Baidu AI Cloud Qianfan ModelBuilder`
+  - Select `API Key` from the left menu
 
-  <Image alt={'Create Application'} inStep src={'https://github.com/user-attachments/assets/927b1040-e23f-4919-92e2-80a400db8327'} />
+  <Image alt={'API Key'} inStep src={'https://github.com/user-attachments/assets/6234428d-5633-4b2f-be22-1a1772a69a55'} />
 
-  - Enter the `Security Authentication` -> `Access Key` management page from the user account menu
-  - Copy the `Access Key` and `Secret Key`, and store them safely
+  - Click `Create API Key`
+  - In `Service`, select `Qianfan ModelBuilder`
+  - In `Resource`, choose `All Resources`
+  - Click the `Confirm` button
+  - Copy the `API Key` and keep it safe
 
-  <Image alt={'Save Keys'} inStep src={'https://github.com/user-attachments/assets/bb9dadd3-7e9c-45fd-8c56-553ab7287453'} />
+  <Image alt={'Create API Key'} inStep src={'https://github.com/user-attachments/assets/6d068fe0-8100-4b43-b0c3-7934f54e688f'} />
+  <Image alt={'Copy API Key'} inStep src={'https://github.com/user-attachments/assets/629adf4e-e9e1-40dc-b9e5-d7b908878170'} />
 
   ### Step 2: Configure Wenxin Qianfan in LobeChat
 
-  - Go to the `Settings` interface in LobeChat
-  - Locate the settings for `Wenxin Qianfan` under `Language Model`
+  - Go to the `Settings` page of LobeChat
+  - Under `Language Models`, find the `Wenxin Qianfan` settings
 
-  <Image alt={'Enter API Keys'} inStep src={'https://github.com/user-attachments/assets/e3995de7-38d9-489b-80a2-434477018469'} />
+  <Image alt={'Enter API Key'} inStep src={'https://github.com/user-attachments/assets/d7666e2a-0202-4b45-8338-9806ddffa44e'} />
 
-  - Enter the obtained `Access Key` and `Secret Key`
-  - Select a Wenxin Qianfan model for your AI assistant to start interacting
+  - Enter the obtained `API Key`
+  - Select a Wenxin Qianfan model for your AI assistant, and you're ready to start chatting!
 
   <Image alt={'Select Wenxin Qianfan Model and Start Chat'} inStep src={'https://github.com/user-attachments/assets/b6e6a3eb-13c6-46f0-9c7c-69a20deae30f'} />
 

package/docs/usage/providers/wenxin.zh-CN.mdx

@@ -22,24 +22,27 @@ tags:
 
   - 注册并登录 [百度智能云控制台](https://console.bce.baidu.com/)
   - 进入 `百度智能云千帆 ModelBuilder`
-  - 在左侧菜单中选择`应用接入`
-  - 创建一个应用
+  - 在左侧菜单中选择 `API Key`
 
-  <Image alt={'创建应用'} inStep src={'https://github.com/user-attachments/assets/927b1040-e23f-4919-92e2-80a400db8327'} />
+  <Image alt={'API Key'} inStep src={'https://github.com/user-attachments/assets/6234428d-5633-4b2f-be22-1a1772a69a55'} />
 
-  - 在用户账号菜单进入 `安全认证` -> `Access Key` 管理页面
-  - 复制 `Access Key` 和 `Secret Key`，并妥善保存
+  - 点击创建 API Key
+  - 在 `服务` 中选择 `千帆ModelBuilder`
+  - 在 `资源` 中选择 `所有资源`
+  - 点击 `确定` 按钮
+  - 复制 `API Key` 并妥善保存
 
-  <Image alt={'保存密钥'} inStep src={'https://github.com/user-attachments/assets/bb9dadd3-7e9c-45fd-8c56-553ab7287453'} />
+  <Image alt={'创建密钥'} inStep src={'https://github.com/user-attachments/assets/6d068fe0-8100-4b43-b0c3-7934f54e688f'} />
+  <Image alt={'复制密钥'} inStep src={'https://github.com/user-attachments/assets/629adf4e-e9e1-40dc-b9e5-d7b908878170'} />
 
   ### 步骤二：在 LobeChat 中配置文心千帆
 
   - 访问 LobeChat 的`设置`界面
   - 在`语言模型`下找到 `文心千帆` 的设置项
 
-  <Image alt={'填入 API 密钥'} inStep src={'https://github.com/user-attachments/assets/e3995de7-38d9-489b-80a2-434477018469'} />
+  <Image alt={'填入 API 密钥'} inStep src={'https://github.com/user-attachments/assets/d7666e2a-0202-4b45-8338-9806ddffa44e'} />
 
-  - 填入获得的 `Access Key` 和 `Secret Key`
+  - 填入获得的 `API Key`
   - 为你的 AI 助手选择一个文心千帆的模型即可开始对话
 
   <Image alt={'选择文心千帆模型并开始对话'} inStep src={'https://github.com/user-attachments/assets/b6e6a3eb-13c6-46f0-9c7c-69a20deae30f'} />

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@lobehub/chat",
-  "version": "1.51.2",
+  "version": "1.51.3",
   "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",
@@ -111,7 +111,6 @@
     "@aws-sdk/s3-request-presigner": "^3.723.0",
     "@azure/core-rest-pipeline": "1.16.0",
     "@azure/openai": "1.0.0-beta.12",
-    "@baiducloud/qianfan": "^0.1.9",
     "@cfworker/json-schema": "^4.1.0",
     "@clerk/localizations": "^3.9.6",
     "@clerk/nextjs": "^6.10.6",

package/src/app/(main)/settings/llm/ProviderList/providers.tsx

@@ -27,6 +27,7 @@ import {
   TaichuProviderCard,
   TogetherAIProviderCard,
   UpstageProviderCard,
+  WenxinProviderCard,
   XAIProviderCard,
   ZeroOneProviderCard,
   ZhiPuProviderCard,
@@ -40,7 +41,6 @@ import { useGithubProvider } from './Github';
 import { useHuggingFaceProvider } from './HuggingFace';
 import { useOllamaProvider } from './Ollama';
 import { useOpenAIProvider } from './OpenAI';
-import { useWenxinProvider } from './Wenxin';
 
 export const useProviderList = (): ProviderItem[] => {
   const AzureProvider = useAzureProvider();
@@ -50,7 +50,6 @@ export const useProviderList = (): ProviderItem[] => {
   const CloudflareProvider = useCloudflareProvider();
   const GithubProvider = useGithubProvider();
   const HuggingFaceProvider = useHuggingFaceProvider();
-  const WenxinProvider = useWenxinProvider();
 
   return useMemo(
     () => [
@@ -75,7 +74,7 @@ export const useProviderList = (): ProviderItem[] => {
       UpstageProviderCard,
       XAIProviderCard,
       QwenProviderCard,
-      WenxinProvider,
+      WenxinProviderCard,
       HunyuanProviderCard,
       SparkProviderCard,
       ZhiPuProviderCard,
@@ -99,7 +98,6 @@ export const useProviderList = (): ProviderItem[] => {
       BedrockProvider,
       CloudflareProvider,
       GithubProvider,
-      WenxinProvider,
       HuggingFaceProvider,
     ],
   );

package/src/config/aiModels/wenxin.ts

@@ -2,12 +2,15 @@ import { AIChatModelCard } from '@/types/aiModel';
 
 const wenxinChatModels: AIChatModelCard[] = [
   {
+    abilities: {
+      functionCall: true,
+    },
     contextWindowTokens: 8192,
     description:
       '百度自研的旗舰级大规模⼤语⾔模型，覆盖海量中英文语料，具有强大的通用能力，可满足绝大部分对话问答、创作生成、插件应用场景要求；支持自动对接百度搜索插件，保障问答信息时效。',
     displayName: 'ERNIE 3.5 8K',
     enabled: true,
-    id: 'ERNIE-3.5-8K',
+    id: 'ernie-3.5-8k',
     pricing: {
       currency: 'CNY',
       input: 0.8,
@@ -16,11 +19,14 @@ const wenxinChatModels: AIChatModelCard[] = [
     type: 'chat',
   },
   {
+    abilities: {
+      functionCall: true,
+    },
     contextWindowTokens: 8192,
     description:
       '百度自研的旗舰级大规模⼤语⾔模型，覆盖海量中英文语料，具有强大的通用能力，可满足绝大部分对话问答、创作生成、插件应用场景要求；支持自动对接百度搜索插件，保障问答信息时效。',
     displayName: 'ERNIE 3.5 8K Preview',
-    id: 'ERNIE-3.5-8K-Preview',
+    id: 'ernie-3.5-8k-preview',
     pricing: {
       currency: 'CNY',
       input: 0.8,
@@ -29,12 +35,15 @@ const wenxinChatModels: AIChatModelCard[] = [
     type: 'chat',
   },
   {
+    abilities: {
+      functionCall: true,
+    },
     contextWindowTokens: 128_000,
     description:
       '百度自研的旗舰级大规模⼤语⾔模型，覆盖海量中英文语料，具有强大的通用能力，可满足绝大部分对话问答、创作生成、插件应用场景要求；支持自动对接百度搜索插件，保障问答信息时效。',
     displayName: 'ERNIE 3.5 128K',
     enabled: true,
-    id: 'ERNIE-3.5-128K',
+    id: 'ernie-3.5-128k',
     pricing: {
       currency: 'CNY',
       input: 0.8,
@@ -43,12 +52,15 @@ const wenxinChatModels: AIChatModelCard[] = [
     type: 'chat',
   },
   {
+    abilities: {
+      functionCall: true,
+    },
     contextWindowTokens: 8192,
     description:
       '百度自研的旗舰级超大规模⼤语⾔模型，相较ERNIE 3.5实现了模型能力全面升级，广泛适用于各领域复杂任务场景；支持自动对接百度搜索插件，保障问答信息时效。',
     displayName: 'ERNIE 4.0 8K',
     enabled: true,
-    id: 'ERNIE-4.0-8K-Latest',
+    id: 'ernie-4.0-8k-latest',
     pricing: {
       currency: 'CNY',
       input: 30,
@@ -57,11 +69,14 @@ const wenxinChatModels: AIChatModelCard[] = [
     type: 'chat',
   },
   {
+    abilities: {
+      functionCall: true,
+    },
     contextWindowTokens: 8192,
     description:
       '百度自研的旗舰级超大规模⼤语⾔模型，相较ERNIE 3.5实现了模型能力全面升级，广泛适用于各领域复杂任务场景；支持自动对接百度搜索插件，保障问答信息时效。',
     displayName: 'ERNIE 4.0 8K Preview',
-    id: 'ERNIE-4.0-8K-Preview',
+    id: 'ernie-4.0-8k-preview',
     pricing: {
       currency: 'CNY',
       input: 30,
@@ -70,12 +85,15 @@ const wenxinChatModels: AIChatModelCard[] = [
     type: 'chat',
   },
   {
+    abilities: {
+      functionCall: true,
+    },
     contextWindowTokens: 8192,
     description:
       '百度自研的旗舰级超大规模⼤语⾔模型，综合效果表现出色，广泛适用于各领域复杂任务场景；支持自动对接百度搜索插件，保障问答信息时效。相较于ERNIE 4.0在性能表现上更优秀',
     displayName: 'ERNIE 4.0 Turbo 8K',
     enabled: true,
-    id: 'ERNIE-4.0-Turbo-8K-Latest',
+    id: 'ernie-4.0-turbo-8k-latest',
     pricing: {
       currency: 'CNY',
       input: 20,
@@ -84,12 +102,15 @@ const wenxinChatModels: AIChatModelCard[] = [
     type: 'chat',
   },
   {
+    abilities: {
+      functionCall: true,
+    },
     contextWindowTokens: 128_000,
     description:
       '百度自研的旗舰级超大规模⼤语⾔模型，综合效果表现出色，广泛适用于各领域复杂任务场景；支持自动对接百度搜索插件，保障问答信息时效。相较于ERNIE 4.0在性能表现上更优秀',
     displayName: 'ERNIE 4.0 Turbo 128K',
     enabled: true,
-    id: 'ERNIE-4.0-Turbo-128K',
+    id: 'ernie-4.0-turbo-128k',
     pricing: {
       currency: 'CNY',
       input: 20,
@@ -98,11 +119,14 @@ const wenxinChatModels: AIChatModelCard[] = [
     type: 'chat',
   },
   {
+    abilities: {
+      functionCall: true,
+    },
     contextWindowTokens: 8192,
     description:
       '百度自研的旗舰级超大规模⼤语⾔模型，综合效果表现出色，广泛适用于各领域复杂任务场景；支持自动对接百度搜索插件，保障问答信息时效。相较于ERNIE 4.0在性能表现上更优秀',
     displayName: 'ERNIE 4.0 Turbo 8K Preview',
-    id: 'ERNIE-4.0-Turbo-8K-Preview',
+    id: 'ernie-4.0-turbo-8k-preview',
     pricing: {
       currency: 'CNY',
       input: 20,
@@ -111,12 +135,27 @@ const wenxinChatModels: AIChatModelCard[] = [
     type: 'chat',
   },
   {
+    contextWindowTokens: 8192,
+    description:
+      'ERNIE Lite是百度自研的轻量级大语言模型，兼顾优异的模型效果与推理性能，适合低算力AI加速卡推理使用。',
+    displayName: 'ERNIE Lite 8K',
+    id: 'ernie-lite-8k',
+    pricing: {
+      currency: 'CNY',
+      input: 0,
+      output: 0,
+    },
+    type: 'chat',
+  },
+  {
+    abilities: {
+      functionCall: true,
+    },
     contextWindowTokens: 128_000,
     description:
       '百度自研的轻量级大语言模型，兼顾优异的模型效果与推理性能，效果比ERNIE Lite更优，适合低算力AI加速卡推理使用。',
     displayName: 'ERNIE Lite Pro 128K',
-    enabled: true,
-    id: 'ERNIE-Lite-Pro-128K',
+    id: 'ernie-lite-pro-128k',
     pricing: {
       currency: 'CNY',
       input: 0.2,
@@ -125,16 +164,15 @@ const wenxinChatModels: AIChatModelCard[] = [
     type: 'chat',
   },
   {
-    contextWindowTokens: 128_000,
+    contextWindowTokens: 8192,
     description:
-      '百度2024年最新发布的自研高性能大语言模型，通用能力优异，效果比ERNIE Speed更优，适合作为基座模型进行精调，更好地处理特定场景问题，同时具备极佳的推理性能。',
-    displayName: 'ERNIE Speed Pro 128K',
-    enabled: true,
-    id: 'ERNIE-Speed-Pro-128K',
+      'ERNIE Tiny是百度自研的超高性能大语言模型，部署与精调成本在文心系列模型中最低。',
+    displayName: 'ERNIE Tiny 8K',
+    id: 'ernie-tiny-8k',
     pricing: {
       currency: 'CNY',
-      input: 0.3,
-      output: 0.6,
+      input: 0,
+      output: 0,
     },
     type: 'chat',
   },
@@ -143,7 +181,7 @@ const wenxinChatModels: AIChatModelCard[] = [
     description:
       '百度2024年最新发布的自研高性能大语言模型，通用能力优异，适合作为基座模型进行精调，更好地处理特定场景问题，同时具备极佳的推理性能。',
     displayName: 'ERNIE Speed 128K',
-    id: 'ERNIE-Speed-128K',
+    id: 'ernie-speed-128k',
     pricing: {
       currency: 'CNY',
       input: 0,
@@ -151,12 +189,25 @@ const wenxinChatModels: AIChatModelCard[] = [
     },
     type: 'chat',
   },
+  {
+    contextWindowTokens: 128_000,
+    description:
+      '百度2024年最新发布的自研高性能大语言模型，通用能力优异，效果比ERNIE Speed更优，适合作为基座模型进行精调，更好地处理特定场景问题，同时具备极佳的推理性能。',
+    displayName: 'ERNIE Speed Pro 128K',
+    id: 'ernie-speed-pro-128k',
+    pricing: {
+      currency: 'CNY',
+      input: 0.3,
+      output: 0.6,
+    },
+    type: 'chat',
+  },
   {
     contextWindowTokens: 8192,
     description:
       '百度自研的垂直场景大语言模型，适合游戏NPC、客服对话、对话角色扮演等应用场景，人设风格更为鲜明、一致，指令遵循能力更强，推理性能更优。',
     displayName: 'ERNIE Character 8K',
-    id: 'ERNIE-Character-8K',
+    id: 'ernie-char-8k',
     pricing: {
       currency: 'CNY',
       input: 4,
@@ -164,6 +215,61 @@ const wenxinChatModels: AIChatModelCard[] = [
     },
     type: 'chat',
   },
+  {
+    contextWindowTokens: 8192,
+    description:
+      '百度自研的垂直场景大语言模型，适合游戏NPC、客服对话、对话角色扮演等应用场景，人设风格更为鲜明、一致，指令遵循能力更强，推理性能更优。',
+    displayName: 'ERNIE Character Fiction 8K',
+    id: 'ernie-char-fiction-8k',
+    pricing: {
+      currency: 'CNY',
+      input: 4,
+      output: 8,
+    },
+    type: 'chat',
+  },
+  {
+    contextWindowTokens: 8192,
+    description:
+      '百度自研通用大语言模型，在小说续写能力上有明显优势，也可用在短剧、电影等场景。',
+    displayName: 'ERNIE Novel 8K',
+    id: 'ernie-novel-8k',
+    pricing: {
+      currency: 'CNY',
+      input: 40,
+      output: 120,
+    },
+    type: 'chat',
+  },
+  {
+    contextWindowTokens: 65_536,
+    description:
+      'DeepSeek-V3 为杭州深度求索人工智能基础技术研究有限公司自研的 MoE 模型，其多项评测成绩突出，在主流榜单中位列开源模型榜首。V3 相比 V2.5 模型生成速度实现 3 倍提升，为用户带来更加迅速流畅的使用体验。',
+    displayName: 'DeepSeek V3',
+    id: 'deepseek-v3',
+    pricing: {
+      currency: 'CNY',
+      input: 0.8,
+      output: 1.6,
+    },
+    type: 'chat',
+  },
+  {
+    abilities: {
+      reasoning: true,
+    },
+    contextWindowTokens: 65_536,
+    description:
+      'DeepSeek-R1 在后训练阶段大规模使用了强化学习技术，在仅有极少标注数据的情况下，极大提升了模型推理能力。在数学、代码、自然语言推理等任务上，性能比肩 OpenAI o1 正式版。',
+    displayName: 'DeepSeek R1',
+    id: 'deepseek-r1',
+    pricing: {
+      currency: 'CNY',
+      input: 2,
+      output: 8,
+    },
+    type: 'chat',
+  },
 ];
 
 export const allModels = [...wenxinChatModels];

package/src/config/llm.ts

@@ -64,8 +64,7 @@ export const getLLMConfig = () => {
       AWS_SESSION_TOKEN: z.string().optional(),
 
       ENABLED_WENXIN: z.boolean(),
-      WENXIN_ACCESS_KEY: z.string().optional(),
-      WENXIN_SECRET_KEY: z.string().optional(),
+      WENXIN_API_KEY: z.string().optional(),
 
       ENABLED_OLLAMA: z.boolean(),
 
@@ -186,9 +185,8 @@ export const getLLMConfig = () => {
       AWS_SECRET_ACCESS_KEY: process.env.AWS_SECRET_ACCESS_KEY,
       AWS_SESSION_TOKEN: process.env.AWS_SESSION_TOKEN,
 
-      ENABLED_WENXIN: !!process.env.WENXIN_ACCESS_KEY && !!process.env.WENXIN_SECRET_KEY,
-      WENXIN_ACCESS_KEY: process.env.WENXIN_ACCESS_KEY,
-      WENXIN_SECRET_KEY: process.env.WENXIN_SECRET_KEY,
+      ENABLED_WENXIN: !!process.env.WENXIN_API_KEY,
+      WENXIN_API_KEY: process.env.WENXIN_API_KEY,
 
       ENABLED_OLLAMA: process.env.ENABLED_OLLAMA !== '0',