issue: 多问题汇总

[Amamiya23]

Check Existing Issues

• I have searched the existing issues and discussions.
• I am using the latest version of Open WebUI.

Installation Method

Git Clone

Open WebUI Version

v0.1

Ollama Version (if applicable)

`

Operating System

`

Browser (if applicable)

`

Confirmation

• I have read and followed all instructions in README.md.
• I am using the latest version of both Open WebUI and Ollama.
• I have included the browser console logs.
• I have included the Docker container logs.
• I have listed steps to reproduce the bug in detail.

Expected Behavior

openwebui原本openai接口可以改为azure的接口，原本是可用的，现在不可用，

Actual Behavior

·

Steps to Reproduce

·

Logs & Screenshots

·

Additional Information

·

[Amamiya23]

还有一个bug，原版openwebui的标题生成ai生成完毕对话后，会自动在左侧显示生成的标题。现在必须刷新才会显示生成的标题

[Amamiya23]

还有一个bug，在设置中设置mcp的标题和描述后，实际并不会在主页面同步，如图2

[Amamiya23]

图像生成这里，希望能够像原版一样能手动输入模型名称，因为大部分时候我使用的是兼容openai,genai sdk的api，此时无法获取模型名称，就无法进行生图功能。

[ztx888]

openwebui原本openai接口可以改为azure的接口，原本是可用的，现在不可用

此问题已定位并修复

[ztx888]

图像生成这里，希望能够像原版一样能手动输入模型名称，因为大部分时候我使用的是兼容openai,genai sdk的api，此时无法获取模型名称，就无法进行生图功能。

图片问题也修复了，之前是个BUG，无法获取模型列表，直接选择即可

[Amamiya23]

图像生成这里，希望能够像原版一样能手动输入模型名称，因为大部分时候我使用的是兼容openai,genai sdk的api，此时无法获取模型名称，就无法进行生图功能。

图片问题也修复了，之前是个BUG，无法获取模型列表，直接选择即可

还没有推送吗，我拉取docker发现似乎没改动

[ztx888]

还有一个bug，原版openwebui的标题生成ai生成完毕对话后，会自动在左侧显示生成的标题。现在必须刷新才会显示生成的标题

还没推送，正在排查其他问题，这个问题我这里没有复现，但猜测可能是生成的比较慢，或者遇到模型是思考模型，在你刷新的那一刻刚好它也生成好了，下次你可以试一下，等待看看会不会自动生成。我后续会优化默认的提示词，尽可能速度更快，这个会在后续版本优化

[ztx888]

现已推送更新，有任何问题欢迎持续反馈

[AmosZhang2025]

MinerU 使用过程中 APIkey 是必填项 ，MinerU 官网给的API接口好像只需要token用来确认用户。不知道应该将官网的token填在哪里。
import requests

token = "官网申请的api token"
url = "https://mineru.net/api/v4/extract/task"
header = {
"Content-Type": "application/json",
"Authorization": f"Bearer {token}"
}
data = {
"url": "https://cdn-mineru.openxlab.org.cn/demo/example.pdf",
"model_version": "vlm"
}

res = requests.post(url,headers=header,json=data)
print(res.status_code)
print(res.json())
print(res.json()["data"])

[ssxwcz]

还有一个bug，原版openwebui的标题生成ai生成完毕对话后，会自动在左侧显示生成的标题。现在必须刷新才会显示生成的标题

还没推送，正在排查其他问题，这个问题我这里没有复现，但猜测可能是生成的比较慢，或者遇到模型是思考模型，在你刷新的那一刻刚好它也生成好了，下次你可以试一下，等待看看会不会自动生成。我后续会优化默认的提示词，尽可能速度更快，这个会在后续版本优化

关于标题生成问题，我这边是直接就没有生成了，看了下右边的请求也都成功了

[ssxwcz]

还有一个bug，原版openwebui的标题生成ai生成完毕对话后，会自动在左侧显示生成的标题。现在必须刷新才会显示生成的标题

还没推送，正在排查其他问题，这个问题我这里没有复现，但猜测可能是生成的比较慢，或者遇到模型是思考模型，在你刷新的那一刻刚好它也生成好了，下次你可以试一下，等待看看会不会自动生成。我后续会优化默认的提示词，尽可能速度更快，这个会在后续版本优化

关于标题生成问题，我这边是直接就没有生成了，看了下右边的请求也都成功了

这是详细的请求日志，我这次换成了gemini-3-flash-preview来做标题生成和生成后续问题，看了下请求日志，这里响应是200了，但响应内容为空。

halowebui  | 2026-04-06 12:04:50.306 | INFO     | open_webui.routers.gemini:_open_gemini_response:864 - Gemini Chat Response Status: 200 - {}
halowebui  | 2026-04-06 12:04:50.306 | DEBUG    | open_webui.routers.tasks:generate_chat_tags:254 - generating chat tags using model gemini-3-flash-preview for user ssxwcz@gmail.com  - {}
halowebui  | 2026-04-06 12:04:50.307 | DEBUG    | open_webui.utils.chat:generate_chat_completion:183 - generate_chat_completion summary={'model': 'gemini-3-flash-preview', 'stream': False, 'message_count': 1, 'file_count': 0} - {}
halowebui  | 2026-04-06 12:04:50.308 | INFO     | open_webui.routers.gemini:generate_chat_completion:1615 - [GEMINI MESSAGES DEBUG] Received 1 messages from middleware - {}
halowebui  | 2026-04-06 12:04:50.308 | INFO     | open_webui.routers.gemini:generate_chat_completion:1621 - [GEMINI MESSAGES DEBUG] Message 0: role=user, has_tool_calls=False, tool_call_id=, content_preview=### Task:
halowebui  | Generate 1-3 broad tags categorizing the main themes of the chat history, along with 1-3 m - {}
halowebui  | 2026-04-06 12:04:50.308 | INFO     | open_webui.routers.gemini:generate_chat_completion:1937 - [GEMINI PAYLOAD DEBUG] Complete contents count: 1 - {}
halowebui  | 2026-04-06 12:04:50.308 | INFO     | open_webui.routers.gemini:generate_chat_completion:1939 - [GEMINI PAYLOAD DEBUG] Content 0: role=user, parts_count=1, parts_types=[['text']] - {}
halowebui  | 2026-04-06 12:04:50.309 | INFO     | open_webui.routers.gemini:generate_chat_completion:1940 - [GEMINI PAYLOAD DEBUG] Complete contents array: [{"role": "user", "parts": [{"text": "### Task:\nGenerate 1-3 broad tags categorizing the main themes of the chat history, along with 1-3 more specific subtopic tags.\n\n### Guidelines:\n- Start with high-level domains (e.g. Science, Technology, Philosophy, Arts, Politics, Business, Health, Sports, Entertainment, Education)\n- Consider including relevant subfields/subdomains if they are strongly represented throughout the conversation\n- If content is too short (less than 3 messages) or too diverse, use only [\"General\"]\n- Use the chat's primary language; default to English if multilingual\n- Prioritize accuracy over specificity\n\n### Output:\nJSON format: { \"tags\": [\"tag1\", \"tag2\", \"tag3\"] }\n\n### Chat History:\n<chat_history>\nUSER: 看下mcp支持那些功能\nASSISTANT: # MCP (Model Context Protocol) 支持的功能\n\nMCP 是一个开放协议，用于让 AI 模型与外部工具和数据源连接。主要支持以下功能：\n\n## 🔌 核心功能\n\n| 功能类别 | 说明 |\n|---------|------|\n| **文件系统** | 读取、写入、搜索本地文件 |\n| **数据库** | 连接 SQL 数据库（PostgreSQL、MySQL 等） |\n| **API 调用** | 与 REST API 交互 |\n| **搜索** | 网络搜索、本地文件搜索 |\n| **执行命令** | 在系统上运行 shell 命令 |\n\n## 🛠️ 常见集成\n\n- **开发工具**: GitHub、Git、Slack、Notion\n- **数据源**: Google Sheets、各类数据库\n- **搜索**: Brave Search、Fetch 内容\n\n## 📦 主要特点\n\n1. **标准化** - 统一的协议规范\n2. **安全** - 支持认证和权限控制\n3. **可扩展** - 开发者可自定义 MCP 服务器\n4. **双向通信** - AI 可主动调用工具并获取结果\n\n## 🏠 官方服务器\n\nAnthropic 提供了几个官方 MCP 服务器：\n- `filesystem` - 文件系统访问\n- `brave-search` - 搜索\n- `fetch` - 获取网页内容\n- `sqlite` - SQLite 数据库\n\n---\n\n如果你想了解某个具体功能或集成，可以告诉我！😊\n</chat_history>"}]}] - {}
halowebui  | 2026-04-06 12:04:50.309 | INFO     | open_webui.routers.gemini:generate_chat_completion:1947 - [GEMINI PAYLOAD DEBUG] Full payload: {"contents": [{"role": "user", "parts": [{"text": "### Task:\nGenerate 1-3 broad tags categorizing the main themes of the chat history, along with 1-3 more specific subtopic tags.\n\n### Guidelines:\n- Start with high-level domains (e.g. Science, Technology, Philosophy, Arts, Politics, Business, Health, Sports, Entertainment, Education)\n- Consider including relevant subfields/subdomains if they are strongly represented throughout the conversation\n- If content is too short (less than 3 messages) or too diverse, use only [\"General\"]\n- Use the chat's primary language; default to English if multilingual\n- Prioritize accuracy over specificity\n\n### Output:\nJSON format: { \"tags\": [\"tag1\", \"tag2\", \"tag3\"] }\n\n### Chat History:\n<chat_history>\nUSER: 看下mcp支持那些功能\nASSISTANT: # MCP (Model Context Protocol) 支持的功能\n\nMCP 是一个开放协议，用于让 AI 模型与外部工具和数据源连接。主要支持以下功能：\n\n## 🔌 核心功能\n\n| 功能类别 | 说明 |\n|---------|------|\n| **文件系统** | 读取、写入、搜索本地文件 |\n| **数据库** | 连接 SQL 数据库（PostgreSQL、MySQL 等） |\n| **API 调用** | 与 REST API 交互 |\n| **搜索** | 网络搜索、本地文件搜索 |\n| **执行命令** | 在系统上运行 shell 命令 |\n\n## 🛠️ 常见集成\n\n- **开发工具**: GitHub、Git、Slack、Notion\n- **数据源**: Google Sheets、各类数据库\n- **搜索**: Brave Search、Fetch 内容\n\n## 📦 主要特点\n\n1. **标准化** - 统一的协议规范\n2. **安全** - 支持认证和权限控制\n3. **可扩展** - 开发者可自定义 MCP 服务器\n4. **双向通信** - AI 可主动调用工具并获取结果\n\n## 🏠 官方服务器\n\nAnthropic 提供了几个官方 MCP 服务器：\n- `filesystem` - 文件系统访问\n- `brave-search` - 搜索\n- `fetch` - 获取网页内容\n- `sqlite` - SQLite 数据库\n\n---\n\n如果你想了解某个具体功能或集成，可以告诉我！😊\n</chat_history>"}]}]} - {}
halowebui  | 2026-04-06 12:04:50.309 | INFO     | open_webui.routers.gemini:generate_chat_completion:1958 - Gemini Chat Request: URL=https://generativelanguage.googleapis.com/v1beta/models/gemini-3-flash-preview:generateContent, AuthType=x-goog-api-key, Stream=False - {}

halowebui  | 2026-04-06 12:04:55.851 | INFO     | open_webui.routers.gemini:_open_gemini_response:864 - Gemini Chat Response Status: 200 - {}
halowebui  | 2026-04-06 12:04:55.851 | DEBUG    | open_webui.routers.tasks:generate_follow_ups:622 - generating follow-ups using model gemini-3-flash-preview for user ssxwcz@gmail.com - {}
halowebui  | 2026-04-06 12:04:55.851 | DEBUG    | open_webui.utils.chat:generate_chat_completion:183 - generate_chat_completion summary={'model': 'gemini-3-flash-preview', 'stream': False, 'message_count': 1, 'file_count': 0} - {}
halowebui  | 2026-04-06 12:04:55.853 | INFO     | open_webui.routers.gemini:generate_chat_completion:1615 - [GEMINI MESSAGES DEBUG] Received 1 messages from middleware - {}
halowebui  | 2026-04-06 12:04:55.853 | INFO     | open_webui.routers.gemini:generate_chat_completion:1621 - [GEMINI MESSAGES DEBUG] Message 0: role=user, has_tool_calls=False, tool_call_id=, content_preview=### Task:
halowebui  | Based on the chat history, generate 3 concise follow-up questions that the user might want - {}
halowebui  | 2026-04-06 12:04:55.853 | INFO     | open_webui.routers.gemini:generate_chat_completion:1937 - [GEMINI PAYLOAD DEBUG] Complete contents count: 1 - {}
halowebui  | 2026-04-06 12:04:55.854 | INFO     | open_webui.routers.gemini:generate_chat_completion:1939 - [GEMINI PAYLOAD DEBUG] Content 0: role=user, parts_count=1, parts_types=[['text']] - {}
halowebui  | 2026-04-06 12:04:55.854 | INFO     | open_webui.routers.gemini:generate_chat_completion:1940 - [GEMINI PAYLOAD DEBUG] Complete contents array: [{"role": "user", "parts": [{"text": "### Task:\nBased on the chat history, generate 3 concise follow-up questions that the user might want to ask next.\n\n### Guidelines:\n- Questions should be relevant to the conversation topic and naturally continue the discussion.\n- Each question should explore a different aspect or direction.\n- Keep questions concise (under 20 words each).\n- Use the same language as the conversation.\n- Do NOT repeat questions already asked.\n\n### Output:\nRespond **EXCLUSIVELY** with a JSON object. Do NOT include any additional text, explanation, or formatting.\nFormat: { \"follow_ups\": [\"question1\", \"question2\", \"question3\"] }\n\n### Chat History:\n<chat_history>\nUSER: 看下mcp支持那些功能\nASSISTANT: # MCP (Model Context Protocol) 支持的功能\n\nMCP 是一个开放协议，用于让 AI 模型与外部工具和数据源连接。主要支持以下功能：\n\n## 🔌 核心功能\n\n| 功能类别 | 说明 |\n|---------|------|\n| **文件系统** | 读取、写入、搜索本地文件 |\n| **数据库** | 连接 SQL 数据库（PostgreSQL、MySQL 等） |\n| **API 调用** | 与 REST API 交互 |\n| **搜索** | 网络搜索、本地文件搜索 |\n| **执行命令** | 在系统上运行 shell 命令 |\n\n## 🛠️ 常见集成\n\n- **开发工具**: GitHub、Git、Slack、Notion\n- **数据源**: Google Sheets、各类数据库\n- **搜索**: Brave Search、Fetch 内容\n\n## 📦 主要特点\n\n1. **标准化** - 统一的协议规范\n2. **安全** - 支持认证和权限控制\n3. **可扩展** - 开发者可自定义 MCP 服务器\n4. **双向通信** - AI 可主动调用工具并获取结果\n\n## 🏠 官方服务器\n\nAnthropic 提供了几个官方 MCP 服务器：\n- `filesystem` - 文件系统访问\n- `brave-search` - 搜索\n- `fetch` - 获取网页内容\n- `sqlite` - SQLite 数据库\n\n---\n\n如果你想了解某个具体功能或集成，可以告诉我！😊\n</chat_history>"}]}] - {}
halowebui  | 2026-04-06 12:04:55.854 | INFO     | open_webui.routers.gemini:generate_chat_completion:1947 - [GEMINI PAYLOAD DEBUG] Full payload: {"contents": [{"role": "user", "parts": [{"text": "### Task:\nBased on the chat history, generate 3 concise follow-up questions that the user might want to ask next.\n\n### Guidelines:\n- Questions should be relevant to the conversation topic and naturally continue the discussion.\n- Each question should explore a different aspect or direction.\n- Keep questions concise (under 20 words each).\n- Use the same language as the conversation.\n- Do NOT repeat questions already asked.\n\n### Output:\nRespond **EXCLUSIVELY** with a JSON object. Do NOT include any additional text, explanation, or formatting.\nFormat: { \"follow_ups\": [\"question1\", \"question2\", \"question3\"] }\n\n### Chat History:\n<chat_history>\nUSER: 看下mcp支持那些功能\nASSISTANT: # MCP (Model Context Protocol) 支持的功能\n\nMCP 是一个开放协议，用于让 AI 模型与外部工具和数据源连接。主要支持以下功能：\n\n## 🔌 核心功能\n\n| 功能类别 | 说明 |\n|---------|------|\n| **文件系统** | 读取、写入、搜索本地文件 |\n| **数据库** | 连接 SQL 数据库（PostgreSQL、MySQL 等） |\n| **API 调用** | 与 REST API 交互 |\n| **搜索** | 网络搜索、本地文件搜索 |\n| **执行命令** | 在系统上运行 shell 命令 |\n\n## 🛠️ 常见集成\n\n- **开发工具**: GitHub、Git、Slack、Notion\n- **数据源**: Google Sheets、各类数据库\n- **搜索**: Brave Search、Fetch 内容\n\n## 📦 主要特点\n\n1. **标准化** - 统一的协议规范\n2. **安全** - 支持认证和权限控制\n3. **可扩展** - 开发者可自定义 MCP 服务器\n4. **双向通信** - AI 可主动调用工具并获取结果\n\n## 🏠 官方服务器\n\nAnthropic 提供了几个官方 MCP 服务器：\n- `filesystem` - 文件系统访问\n- `brave-search` - 搜索\n- `fetch` - 获取网页内容\n- `sqlite` - SQLite 数据库\n\n---\n\n如果你想了解某个具体功能或集成，可以告诉我！😊\n</chat_history>"}]}]} - {}
halowebui  | 2026-04-06 12:04:55.854 | INFO     | open_webui.routers.gemini:generate_chat_completion:1958 - Gemini Chat Request: URL=https://generativelanguage.googleapis.com/v1beta/models/gemini-3-flash-preview:generateContent, AuthType=x-goog-api-key, Stream=False - {}
halowebui  | 2026-04-06 12:04:59.886 | INFO     | open_webui.routers.gemini:_open_gemini_response:864 - Gemini Chat Response Status: 200 - {}

我换成openai格式的模型测试的请求日志就正常了，响应内容中也输出了标题名，这个功能应该是不支持Gemini格式的api吧，建议支持下Gemini格式或者加个说明提醒下用户。

halowebui  | 2026-04-06 12:13:40.057 | INFO     | open_webui.routers.openai:_send_current_request:2090 - [UPSTREAM RESPONSE] status=200 | content-type=application/json - {}
halowebui  | 2026-04-06 12:13:40.058 | DEBUG    | open_webui.routers.openai:_send_current_request:2097 - [UPSTREAM RESPONSE] headers={'Content-Type': 'application/json', 'Content-Length': '2058', 'Date': 'Mon, 06 Apr 2026 12:13:39 GMT', 'req-cost-time': '5676', 'req-arrive-time': '1775477614370', 'resp-start-time': '1775477620046', 'x-envoy-upstream-service-time': '5672', 'Server': 'istio-envoy'} - {}
halowebui  | 2026-04-06 12:13:40.058 | DEBUG    | open_webui.routers.openai:generate_chat_completion:2439 - [UPSTREAM RESPONSE BODY] {"created": 1775477620, "usage": {"completion_tokens": 366, "prompt_tokens": 538, "completion_tokens_details": {"reasoning_tokens": 343}, "prompt_tokens_details": {"cached_tokens": 0}, "total_tokens": 904}, "model": "MiniMax-M2.5", "id": "chatcmpl-1ce574c6-82f8-93ff-acc2-72de58037a7d", "choices": [{"finish_reason": "stop", "index": 0, "message": {"role": "assistant", "content": "```json\n{\n  \"title\": \"🔧 MCP协议功能详解\"\n}\n```", "reasoning_content": "The user is asking me to generate a concise 3-5 word title with an emoji summarizing the chat history.\n\nLooking at the chat history:\n- User asked: \"看下mcp支持那些功能\" (Look at what functions MCP supports)\n- Assistant provided a detailed overview of MCP (Model Context Protocol) features including:\n  - Core functions (Tools, Resources, Prompts, Sampling)\n  - Architecture capabilities (bidirectional communication, auto-discovery, standardized interfaces, security sandbox)\n  - Common integrations (file system, database, Web API, development tools, search)\n  - Workflow diagram\n\nThe conversation is entirely in Chinese (Simplified Chinese), so the title should be in Chinese.\n\nThe main topic is about MCP (Model Context Protocol) and its supported features/capabilities.\n\nPossible titles:\n- \"MCP功能支持概述\" (MCP Feature Support Overview)\n- \"🔧 MCP协议功能详解\" (MCP Protocol Features Explained)\n- \"MCP核心功能介绍\" (MCP Core Features Introduction)\n- \"MCP工具与资源功能\" (MCP Tools and Resources Features)\n\nI think a good title would be: \" MCP功能支持概述\" or \"MCP协议功能详解\"\n\nLet me count words:\n- \"MCP功能支持概述\" = 5 words (MCP/功能/支持/概述)\n- \" MCP协议功能详解\" = 5 words (MCP/协议/功能/详解) plus emoji\n\nThe examples show they use emojis when appropriate. Since the topic is about technical features/capabilities of MCP, the  emoji would be appropriate.\n\nLet me go with: \" MCP协议功能详解\""}, "logprobs": null}], "system_fingerprint": null, "object": "chat.completion"} - {}
halowebui  | 2026-04-06 12:13:40.061 | DEBUG    | open_webui.routers.tasks:generate_chat_tags:254 - generating chat tags using model 862224bb.MiniMax-M2.5 for user ssxwcz@gmail.com  - {}
halowebui  | 2026-04-06 12:13:40.061 | DEBUG    | open_webui.utils.chat:generate_chat_completion:183 - generate_chat_completion summary={'model': '862224bb.MiniMax-M2.5', 'stream': False, 'message_count': 1, 'file_count': 0} - {}
halowebui  | 2026-04-06 12:13:40.062 | INFO     | open_webui.routers.openai:generate_chat_completion:2027 - [UPSTREAM REQUEST] POST https://coding.dashscope.aliyuncs.com/v1/chat/completions | model=MiniMax-M2.5 | payload_keys=['messages', 'model', 'stream'] | messages=1 | tools=0 | size=1521 | reasoning=none | store=None | include=None - {}
halowebui  | 2026-04-06 12:13:40.062 | DEBUG    | open_webui.routers.openai:generate_chat_completion:2044 - [UPSTREAM REQUEST] headers={'Authorization': '***', 'Accept': 'application/json', 'Content-Type': 'application/json'} - {}
halowebui  | 2026-04-06 12:13:40.062 | DEBUG    | open_webui.routers.openai:generate_chat_completion:2058 - [UPSTREAM REQUEST] payload={"model": "MiniMax-M2.5", "messages": [{"role": "user", "content": "### Task:\nGenerate 1-3 broad tags categorizing the main themes of the chat history, along with 1-3 more specific subtopic tags.\n\n### Guidelines:\n- Start with high-level domains (e.g. Science, Technolo...[truncated, total 1368 chars]"}], "stream": false} - {}
halowebui  | 2026-04-06 12:13:40.105 | INFO     | uvicorn.protocols.http.httptools_impl:send:476 - 219.133.69.125:0 - "GET /api/v1/chats/?page=1 HTTP/1.1" 200 - {}
halowebui  | 2026-04-06 12:13:45.919 | INFO     | open_webui.routers.openai:_send_current_request:2090 - [UPSTREAM RESPONSE] status=200 | content-type=application/json - {}
halowebui  | 2026-04-06 12:13:45.919 | DEBUG    | open_webui.routers.openai:_send_current_request:2097 - [UPSTREAM RESPONSE] headers={'Content-Type': 'application/json', 'Content-Length': '2669', 'Date': 'Mon, 06 Apr 2026 12:13:45 GMT', 'req-cost-time': '5661', 'req-arrive-time': '1775477620244', 'resp-start-time': '1775477625905', 'x-envoy-upstream-service-time': '5657', 'Server': 'istio-envoy'} - {}
halowebui  | 2026-04-06 12:13:45.920 | DEBUG    | open_webui.routers.openai:generate_chat_completion:2439 - [UPSTREAM RESPONSE BODY] {"created": 1775477626, "usage": {"completion_tokens": 438, "prompt_tokens": 501, "completion_tokens_details": {"reasoning_tokens": 420}, "prompt_tokens_details": {"cached_tokens": 0}, "total_tokens": 939}, "model": "MiniMax-M2.5", "id": "chatcmpl-5efd1183-139f-906c-9f27-6d307cf9dc8e", "choices": [{"finish_reason": "stop", "index": 0, "message": {"role": "assistant", "content": "{\"tags\": [\"科技\", \"人工智能\", \"软件开发\"]}", "reasoning_content": "Let me analyze the chat history to determine the appropriate tags.\n\nThe conversation is in Chinese (Simplified). The user asks about \"mcp支持那些功能\" (what functions does MCP support). The assistant provides a detailed response about MCP (Model Context Protocol), covering:\n- Core functions (tools, resources, prompts, sampling)\n- Architecture capabilities (bidirectional communication, auto-discovery, standardized interfaces, security sandbox)\n- Common integrations (filesystem, database, Web API, development tools, search)\n- Workflow explanation\n\nThis is clearly about:\n1. Technology/Software - specifically about a protocol (MCP - Model Context Protocol)\n2. AI/ML - It's about connecting AI models with external tools\n3. Software Development - mentions APIs, integration, development tools\n\nPrimary themes:\n- Technology (high-level)\n- AI (more specific subfield)\n- Software Architecture/Development (more specific subfield)\n\nThe content is definitely more than 3 messages and is focused on a specific technical topic, not too diverse.\n\nI should use Chinese as the primary language since the conversation is in Chinese.\n\nLet me create appropriate tags:\n- High-level: Technology, AI\n- More specific: Software Architecture, Development Tools\n\nOutput format: {\"tags\": [\"tag1\", \"tag2\", \"tag3\"]}\n\nIn Chinese: 科技, 人工智能, 软件开发\n\nWait, the guidelines say to start with high-level domains like Science, Technology, Philosophy, etc. Let me reconsider:\n\nHigh-level domains represented:\n- Technology (most appropriate)\n\nSub - {}
halowebui  | 2026-04-06 12:13:45.930 | DEBUG    | open_webui.routers.tasks:generate_follow_ups:622 - generating follow-ups using model 862224bb.MiniMax-M2.5 for user ssxwcz@gmail.com - {}
halowebui  | 2026-04-06 12:13:45.930 | DEBUG    | open_webui.utils.chat:generate_chat_completion:183 - generate_chat_completion summary={'model': '862224bb.MiniMax-M2.5', 'stream': False, 'message_count': 1, 'file_count': 0} - {}
halowebui  | 2026-04-06 12:13:45.931 | INFO     | open_webui.routers.openai:generate_chat_completion:2027 - [UPSTREAM REQUEST] POST https://coding.dashscope.aliyuncs.com/v1/chat/completions | model=MiniMax-M2.5 | payload_keys=['max_tokens', 'messages', 'model', 'stream'] | messages=1 | tools=0 | size=1497 | reasoning=none | store=None | include=None - {}
halowebui  | 2026-04-06 12:13:45.931 | DEBUG    | open_webui.routers.openai:generate_chat_completion:2044 - [UPSTREAM REQUEST] headers={'Authorization': '***', 'Accept': 'application/json', 'Content-Type': 'application/json'} - {}
halowebui  | 2026-04-06 12:13:45.931 | DEBUG    | open_webui.routers.openai:generate_chat_completion:2058 - [UPSTREAM REQUEST] payload={"model": "MiniMax-M2.5", "messages": [{"role": "user", "content": "### Task:\nBased on the chat history, generate 3 concise follow-up questions that the user might want to ask next.\n\n### Guidelines:\n- Questions should be relevant to the conversation topic and naturall...[truncated, total 1326 chars]"}], "stream": false, "max_tokens": 256} - {}
halowebui  | 2026-04-06 12:13:45.972 | INFO     | uvicorn.protocols.http.httptools_impl:send:476 - 219.133.69.125:0 - "GET /api/v1/chats/8864475b-30d0-44d3-b36a-f4901999d0d5 HTTP/1.1" 200 - {}
halowebui  | 2026-04-06 12:13:46.015 | INFO     | uvicorn.protocols.http.httptools_impl:send:476 - 219.133.69.125:0 - "GET /api/v1/chats/all/tags HTTP/1.1" 200 - {}
halowebui  | 2026-04-06 12:13:51.159 | INFO     | open_webui.routers.openai:_send_current_request:2090 - [UPSTREAM RESPONSE] status=200 | content-type=application/json - {}
halowebui  | 2026-04-06 12:13:51.160 | DEBUG    | open_webui.routers.openai:_send_current_request:2097 - [UPSTREAM RESPONSE] headers={'Content-Type': 'application/json', 'Content-Length': '1981', 'Date': 'Mon, 06 Apr 2026 12:13:51 GMT', 'req-cost-time': '5033', 'req-arrive-time': '1775477626112', 'resp-start-time': '1775477631146', 'x-envoy-upstream-service-time': '5030', 'Server': 'istio-envoy'} - {}
halowebui  | 2026-04-06 12:13:51.160 | DEBUG    | open_webui.routers.openai:generate_chat_completion:2439 - [UPSTREAM RESPONSE BODY] {"created": 1775477631, "usage": {"completion_tokens": 339, "prompt_tokens": 483, "completion_tokens_details": {"reasoning_tokens": 288}, "prompt_tokens_details": {"cached_tokens": 0}, "total_tokens": 822}, "model": "MiniMax-M2.5", "id": "chatcmpl-687db0c3-be98-973c-9f05-0ee14956b668", "choices": [{"finish_reason": "stop", "index": 0, "message": {"role": "assistant", "content": "```json\n{\n  \"follow_ups\": [\n    \"如何快速搭建一个MCP服务器？\",\n    \"MCP和其他AI协议有什么区别？\",\n    \"MCP有哪些实际应用场景？\"\n  ]\n}\n```", "reasoning_content": "The user asked about what features MCP (Model Context Protocol) supports, and I provided a detailed response covering core features, architecture capabilities, common integrations, and workflow.\n\nNow I need to generate 3 follow-up questions that the user might want to ask next. These should:\n1. Be relevant to the conversation topic (MCP)\n2. Explore different aspects or directions\n3. Be concise (under 20 words)\n4. Use the same language (Chinese)\n5. Not repeat questions already asked\n\nSome possible directions:\n1. How to implement or use MCP (practical usage)\n2. Differences with other protocols\n3. Security aspects\n4. Specific use cases\n5. How to get started / tutorials\n\nGood follow-up questions could be:\n1. \"如何快速搭建一个MCP服务器？\" (How to quickly set up an MCP server?)\n2. \"MCP和其他AI协议有什么区别？\" (What's the difference between MCP and other AI protocols?)\n3. \"MCP有哪些实际应用场景？\" (What are the practical application scenarios for MCP?)\n\nLet me make sure each is under 20 words:\n- \"如何快速搭建一个MCP服务器？\" - 12 words ✓\n- \"MCP和其他AI协议有什么区别？\" - 10 words ✓\n- \"MCP有哪些实际应用场景？\" - 9 words ✓\n\nThese look good and cover different aspects: implementation, comparison, and practical applications."}, "logprobs": null}], "system_fingerprint": null, "object": "chat.completion"} - {}

[Bduoluoluo]

模型的思考在15分钟时必定会终止，点击 继续生成 也不会保留原有思考内容，将会重新思考