Merge 812c1c770c into f7cde17919

feat(alibaba): Added alibaba vision model and omni model support

README.md

@@ -22,7 +22,6 @@ English / [简体中文](./README_CN.md)
 [![MacOS][MacOS-image]][download-url]
 [![Linux][Linux-image]][download-url]
 
-[NextChatAI](https://nextchat.dev/chat?utm_source=readme) / [Web App Demo](https://app.nextchat.dev) / [Desktop App](https://github.com/Yidadaa/ChatGPT-Next-Web/releases) 
 [NextChatAI](https://nextchat.club?utm_source=readme) / [Web App Demo](https://app.nextchat.dev) / [Desktop App](https://github.com/Yidadaa/ChatGPT-Next-Web/releases) / [Discord](https://discord.gg/YCkeafCafC) / [Enterprise Edition](#enterprise-edition) / [Twitter](https://twitter.com/NextChatDev)
 
 
@@ -56,7 +55,7 @@ Will you come and do something together? 😎
 
 https://github.com/ChatGPTNextWeb/NextChat/issues/6269
 
-#Seeking for talents is thirsty #lack of people"
+#Seeking for talents is thirsty #lack of people
 
 
 ## 🥳 Cheer for DeepSeek, China's AI star!

app/client/api.ts

@@ -40,6 +40,11 @@ export interface MultimodalContent {
   };
 }
 
+export interface MultimodalContentForAlibaba {
+  text?: string;
+  image?: string;
+}
+
 export interface RequestMessage {
   role: MessageRole;
   content: string | MultimodalContent[];

app/client/platforms/alibaba.ts

@@ -7,7 +7,10 @@ import {
   ChatMessageTool,
   usePluginStore,
 } from "@/app/store";
-import { streamWithThink } from "@/app/utils/chat";
+import {
+  preProcessImageContentForAlibabaDashScope,
+  streamWithThink,
+} from "@/app/utils/chat";
 import {
   ChatOptions,
   getHeaders,
@@ -15,12 +18,14 @@ import {
   LLMModel,
   SpeechOptions,
   MultimodalContent,
+  MultimodalContentForAlibaba,
 } from "../api";
 import { getClientConfig } from "@/app/config/client";
 import {
   getMessageTextContent,
   getMessageTextContentWithoutThinking,
   getTimeoutMSByModel,
+  isVisionModel,
 } from "@/app/utils";
 import { fetch } from "@/app/utils/stream";
 
@@ -89,14 +94,6 @@ export class QwenApi implements LLMApi {
   }
 
   async chat(options: ChatOptions) {
-    const messages = options.messages.map((v) => ({
-      role: v.role,
-      content:
-        v.role === "assistant"
-          ? getMessageTextContentWithoutThinking(v)
-          : getMessageTextContent(v),
-    }));
-
     const modelConfig = {
       ...useAppConfig.getState().modelConfig,
       ...useChatStore.getState().currentSession().mask.modelConfig,
@@ -105,6 +102,21 @@ export class QwenApi implements LLMApi {
       },
     };
 
+    const visionModel = isVisionModel(options.config.model);
+
+    const messages: ChatOptions["messages"] = [];
+    for (const v of options.messages) {
+      const content = (
+        visionModel
+          ? await preProcessImageContentForAlibabaDashScope(v.content)
+          : v.role === "assistant"
+          ? getMessageTextContentWithoutThinking(v)
+          : getMessageTextContent(v)
+      ) as any;
+
+      messages.push({ role: v.role, content });
+    }
+
     const shouldStream = !!options.config.stream;
     const requestPayload: RequestPayload = {
       model: modelConfig.model,
@@ -129,7 +141,7 @@ export class QwenApi implements LLMApi {
         "X-DashScope-SSE": shouldStream ? "enable" : "disable",
       };
 
-      const chatPath = this.path(Alibaba.ChatPath);
+      const chatPath = this.path(Alibaba.ChatPath(modelConfig.model));
       const chatPayload = {
         method: "POST",
         body: JSON.stringify(requestPayload),
@@ -162,7 +174,7 @@ export class QwenApi implements LLMApi {
             const json = JSON.parse(text);
             const choices = json.output.choices as Array<{
               message: {
-                content: string | null;
+                content: string | null | MultimodalContentForAlibaba[];
                 tool_calls: ChatMessageTool[];
                 reasoning_content: string | null;
               };
@@ -212,7 +224,9 @@ export class QwenApi implements LLMApi {
             } else if (content && content.length > 0) {
               return {
                 isThinking: false,
-                content: content,
+                content: Array.isArray(content)
+                  ? content.map((item) => item.text).join(",")
+                  : content,
               };
             }
 

app/client/platforms/deepseek.ts

@@ -75,6 +75,25 @@ export class DeepSeekApi implements LLMApi {
       }
     }
 
+    // 检测并修复消息顺序，确保除system外的第一个消息是user
+    const filteredMessages: ChatOptions["messages"] = [];
+    let hasFoundFirstUser = false;
+
+    for (const msg of messages) {
+      if (msg.role === "system") {
+        // Keep all system messages
+        filteredMessages.push(msg);
+      } else if (msg.role === "user") {
+        // User message directly added
+        filteredMessages.push(msg);
+        hasFoundFirstUser = true;
+      } else if (hasFoundFirstUser) {
+        // After finding the first user message, all subsequent non-system messages are retained.
+        filteredMessages.push(msg);
+      }
+      // If hasFoundFirstUser is false and it is not a system message, it will be skipped.
+    }
+
     const modelConfig = {
       ...useAppConfig.getState().modelConfig,
       ...useChatStore.getState().currentSession().mask.modelConfig,
@@ -85,7 +104,7 @@ export class DeepSeekApi implements LLMApi {
     };
 
     const requestPayload: RequestPayload = {
-      messages,
+      messages: filteredMessages,
       stream: options.config.stream,
       model: modelConfig.model,
       temperature: modelConfig.temperature,

app/components/emoji.tsx

@@ -66,11 +66,11 @@ export function Avatar(props: { model?: ModelType; avatar?: string }) {
       LlmIcon = BotIconGemma;
     } else if (modelName.startsWith("claude")) {
       LlmIcon = BotIconClaude;
-    } else if (modelName.toLowerCase().includes("llama")) {
+    } else if (modelName.includes("llama")) {
       LlmIcon = BotIconMeta;
-    } else if (modelName.startsWith("mixtral")) {
+    } else if (modelName.startsWith("mixtral") || modelName.startsWith("codestral")) {
       LlmIcon = BotIconMistral;
-    } else if (modelName.toLowerCase().includes("deepseek")) {
+    } else if (modelName.includes("deepseek")) {
       LlmIcon = BotIconDeepseek;
     } else if (modelName.startsWith("moonshot")) {
       LlmIcon = BotIconMoonshot;
@@ -85,7 +85,7 @@ export function Avatar(props: { model?: ModelType; avatar?: string }) {
     } else if (modelName.startsWith("doubao") || modelName.startsWith("ep-")) {
       LlmIcon = BotIconDoubao;
     } else if (
-      modelName.toLowerCase().includes("glm") ||
+      modelName.includes("glm") ||
       modelName.startsWith("cogview-") ||
       modelName.startsWith("cogvideox-")
     ) {

app/constant.ts

@@ -221,7 +221,12 @@ export const ByteDance = {
 
 export const Alibaba = {
   ExampleEndpoint: ALIBABA_BASE_URL,
-  ChatPath: "v1/services/aigc/text-generation/generation",
+  ChatPath: (modelName: string) => {
+    if (modelName.includes("vl") || modelName.includes("omni")) {
+      return "v1/services/aigc/multimodal-generation/generation";
+    }
+    return `v1/services/aigc/text-generation/generation`;
+  },
 };
 
 export const Tencent = {
@@ -535,6 +540,8 @@ const anthropicModels = [
   "claude-3-5-sonnet-20240620",
   "claude-3-5-sonnet-20241022",
   "claude-3-5-sonnet-latest",
+  "claude-3-7-sonnet-20250219",
+  "claude-3-7-sonnet-latest",
 ];
 
 const baiduModels = [
@@ -568,6 +575,9 @@ const alibabaModes = [
   "qwen-max-0403",
   "qwen-max-0107",
   "qwen-max-longcontext",
+  "qwen-omni-turbo",
+  "qwen-vl-plus",
+  "qwen-vl-max",
 ];
 
 const tencentModels = [

app/utils/chat.ts

@@ -3,7 +3,7 @@ import {
   UPLOAD_URL,
   REQUEST_TIMEOUT_MS,
 } from "@/app/constant";
-import { RequestMessage } from "@/app/client/api";
+import { MultimodalContent, RequestMessage } from "@/app/client/api";
 import Locale from "@/app/locales";
 import {
   EventStreamContentType,
@@ -70,8 +70,9 @@ export function compressImage(file: Blob, maxSize: number): Promise<string> {
   });
 }
 
-export async function preProcessImageContent(
+export async function preProcessImageContentBase(
   content: RequestMessage["content"],
+  transformImageUrl: (url: string) => Promise<{ [key: string]: any }>,
 ) {
   if (typeof content === "string") {
     return content;
@@ -81,7 +82,7 @@ export async function preProcessImageContent(
     if (part?.type == "image_url" && part?.image_url?.url) {
       try {
         const url = await cacheImageToBase64Image(part?.image_url?.url);
-        result.push({ type: part.type, image_url: { url } });
+        result.push(await transformImageUrl(url));
       } catch (error) {
         console.error("Error processing image URL:", error);
       }
@@ -92,6 +93,23 @@ export async function preProcessImageContent(
   return result;
 }
 
+export async function preProcessImageContent(
+  content: RequestMessage["content"],
+) {
+  return preProcessImageContentBase(content, async (url) => ({
+    type: "image_url",
+    image_url: { url },
+  })) as Promise<MultimodalContent[] | string>;
+}
+
+export async function preProcessImageContentForAlibabaDashScope(
+  content: RequestMessage["content"],
+) {
+  return preProcessImageContentBase(content, async (url) => ({
+    image: url,
+  }));
+}
+
 const imageCaches: Record<string, string> = {};
 export function cacheImageToBase64Image(imageUrl: string) {
   if (imageUrl.includes(CACHE_URL_PREFIX)) {

public/plugins.json

@@ -2,16 +2,16 @@
   {
     "id": "dalle3",
     "name": "Dalle3",
-    "schema": "https://ghp.ci/https://raw.githubusercontent.com/ChatGPTNextWeb/NextChat-Awesome-Plugins/main/plugins/dalle/openapi.json"
+    "schema": "https://cdn.jsdelivr.net/gh/ChatGPTNextWeb/NextChat-Awesome-Plugins@main/plugins/dalle/openapi.json"
   },
   {
     "id": "arxivsearch",
     "name": "ArxivSearch",
-    "schema": "https://ghp.ci/https://raw.githubusercontent.com/ChatGPTNextWeb/NextChat-Awesome-Plugins/main/plugins/arxivsearch/openapi.json"
+    "schema": "https://cdn.jsdelivr.net/gh/ChatGPTNextWeb/NextChat-Awesome-Plugins@main/plugins/arxivsearch/openapi.json"
   },
   {
     "id": "duckduckgolite",
     "name": "DuckDuckGoLiteSearch",
-    "schema": "https://ghp.ci/https://raw.githubusercontent.com/ChatGPTNextWeb/NextChat-Awesome-Plugins/main/plugins/duckduckgolite/openapi.json"
+    "schema": "https://cdn.jsdelivr.net/gh/ChatGPTNextWeb/NextChat-Awesome-Plugins@main/plugins/duckduckgolite/openapi.json"
   }
 ]

scripts/fetch-prompts.mjs

@@ -1,14 +1,13 @@
 import fetch from "node-fetch";
 import fs from "fs/promises";
 
-const RAW_FILE_URL = "https://raw.githubusercontent.com/";
-const MIRRORF_FILE_URL = "http://raw.fgit.ml/";
+const MIRRORF_FILE_URL = "https://cdn.jsdelivr.net/gh/";
 
-const RAW_CN_URL = "PlexPt/awesome-chatgpt-prompts-zh/main/prompts-zh.json";
+const RAW_CN_URL = "PlexPt/awesome-chatgpt-prompts-zh@main/prompts-zh.json";
 const CN_URL = MIRRORF_FILE_URL + RAW_CN_URL;
-const RAW_TW_URL = "PlexPt/awesome-chatgpt-prompts-zh/main/prompts-zh-TW.json";
+const RAW_TW_URL = "PlexPt/awesome-chatgpt-prompts-zh@main/prompts-zh-TW.json";
 const TW_URL = MIRRORF_FILE_URL + RAW_TW_URL;
-const RAW_EN_URL = "f/awesome-chatgpt-prompts/main/prompts.csv";
+const RAW_EN_URL = "f/awesome-chatgpt-prompts@main/prompts.csv";
 const EN_URL = MIRRORF_FILE_URL + RAW_EN_URL;
 const FILE = "./public/prompts.json";
 
@@ -84,11 +83,11 @@ async function fetchEN() {
 async function main() {
   Promise.all([fetchCN(), fetchTW(), fetchEN()])
     .then(([cn, tw, en]) => {
-      fs.writeFile(FILE, JSON.stringify({ cn, tw, en }));
+      fs.writeFile(FILE, JSON.stringify({ cn, tw, en }, null, 2));
     })
     .catch((e) => {
       console.error("[Fetch] failed to fetch prompts");
-      fs.writeFile(FILE, JSON.stringify({ cn: [], tw: [], en: [] }));
+      fs.writeFile(FILE, JSON.stringify({ cn: [], tw: [], en: [] }, null, 2));
     })
     .finally(() => {
       console.log("[Fetch] saved to " + FILE);