[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-enricoros--big-AGI":3,"tool-enricoros--big-AGI":62},[4,18,28,37,45,53],{"id":5,"name":6,"github_repo":7,"description_zh":8,"stars":9,"difficulty_score":10,"last_commit_at":11,"category_tags":12,"status":17},4358,"openclaw","openclaw\u002Fopenclaw","OpenClaw 是一款专为个人打造的本地化 AI 助手，旨在让你在自己的设备上拥有完全可控的智能伙伴。它打破了传统 AI 助手局限于特定网页或应用的束缚，能够直接接入你日常使用的各类通讯渠道，包括微信、WhatsApp、Telegram、Discord、iMessage 等数十种平台。无论你在哪个聊天软件中发送消息，OpenClaw 都能即时响应，甚至支持在 macOS、iOS 和 Android 设备上进行语音交互，并提供实时的画布渲染功能供你操控。\n\n这款工具主要解决了用户对数据隐私、响应速度以及“始终在线”体验的需求。通过将 AI 部署在本地，用户无需依赖云端服务即可享受快速、私密的智能辅助，真正实现了“你的数据，你做主”。其独特的技术亮点在于强大的网关架构，将控制平面与核心助手分离，确保跨平台通信的流畅性与扩展性。\n\nOpenClaw 非常适合希望构建个性化工作流的技术爱好者、开发者，以及注重隐私保护且不愿被单一生态绑定的普通用户。只要具备基础的终端操作能力（支持 macOS、Linux 及 Windows WSL2），即可通过简单的命令行引导完成部署。如果你渴望拥有一个懂你",349277,3,"2026-04-06T06:32:30",[13,14,15,16],"Agent","开发框架","图像","数据工具","ready",{"id":19,"name":20,"github_repo":21,"description_zh":22,"stars":23,"difficulty_score":24,"last_commit_at":25,"category_tags":26,"status":17},9989,"n8n","n8n-io\u002Fn8n","n8n 是一款面向技术团队的公平代码（fair-code）工作流自动化平台，旨在让用户在享受低代码快速构建便利的同时，保留编写自定义代码的灵活性。它主要解决了传统自动化工具要么过于封闭难以扩展、要么完全依赖手写代码效率低下的痛点，帮助用户轻松连接 400 多种应用与服务，实现复杂业务流程的自动化。\n\nn8n 特别适合开发者、工程师以及具备一定技术背景的业务人员使用。其核心亮点在于“按需编码”：既可以通过直观的可视化界面拖拽节点搭建流程，也能随时插入 JavaScript 或 Python 代码、调用 npm 包来处理复杂逻辑。此外，n8n 原生集成了基于 LangChain 的 AI 能力，支持用户利用自有数据和模型构建智能体工作流。在部署方面，n8n 提供极高的自由度，支持完全自托管以保障数据隐私和控制权，也提供云端服务选项。凭借活跃的社区生态和数百个现成模板，n8n 让构建强大且可控的自动化系统变得简单高效。",184740,2,"2026-04-19T23:22:26",[16,14,13,15,27],"插件",{"id":29,"name":30,"github_repo":31,"description_zh":32,"stars":33,"difficulty_score":10,"last_commit_at":34,"category_tags":35,"status":17},10095,"AutoGPT","Significant-Gravitas\u002FAutoGPT","AutoGPT 是一个旨在让每个人都能轻松使用和构建 AI 的强大平台，核心功能是帮助用户创建、部署和管理能够自动执行复杂任务的连续型 AI 智能体。它解决了传统 AI 应用中需要频繁人工干预、难以自动化长流程工作的痛点，让用户只需设定目标，AI 即可自主规划步骤、调用工具并持续运行直至完成任务。\n\n无论是开发者、研究人员，还是希望提升工作效率的普通用户，都能从 AutoGPT 中受益。开发者可利用其低代码界面快速定制专属智能体；研究人员能基于开源架构探索多智能体协作机制；而非技术背景用户也可直接选用预置的智能体模板，立即投入实际工作场景。\n\nAutoGPT 的技术亮点在于其模块化“积木式”工作流设计——用户通过连接功能块即可构建复杂逻辑，每个块负责单一动作，灵活且易于调试。同时，平台支持本地自托管与云端部署两种模式，兼顾数据隐私与使用便捷性。配合完善的文档和一键安装脚本，即使是初次接触的用户也能在几分钟内启动自己的第一个 AI 智能体。AutoGPT 正致力于降低 AI 应用门槛，让人人都能成为 AI 的创造者与受益者。",183572,"2026-04-20T04:47:55",[13,36,27,14,15],"语言模型",{"id":38,"name":39,"github_repo":40,"description_zh":41,"stars":42,"difficulty_score":10,"last_commit_at":43,"category_tags":44,"status":17},3808,"stable-diffusion-webui","AUTOMATIC1111\u002Fstable-diffusion-webui","stable-diffusion-webui 是一个基于 Gradio 构建的网页版操作界面，旨在让用户能够轻松地在本地运行和使用强大的 Stable Diffusion 图像生成模型。它解决了原始模型依赖命令行、操作门槛高且功能分散的痛点，将复杂的 AI 绘图流程整合进一个直观易用的图形化平台。\n\n无论是希望快速上手的普通创作者、需要精细控制画面细节的设计师，还是想要深入探索模型潜力的开发者与研究人员，都能从中获益。其核心亮点在于极高的功能丰富度：不仅支持文生图、图生图、局部重绘（Inpainting）和外绘（Outpainting）等基础模式，还独创了注意力机制调整、提示词矩阵、负向提示词以及“高清修复”等高级功能。此外，它内置了 GFPGAN 和 CodeFormer 等人脸修复工具，支持多种神经网络放大算法，并允许用户通过插件系统无限扩展能力。即使是显存有限的设备，stable-diffusion-webui 也提供了相应的优化选项，让高质量的 AI 艺术创作变得触手可及。",162132,"2026-04-05T11:01:52",[14,15,13],{"id":46,"name":47,"github_repo":48,"description_zh":49,"stars":50,"difficulty_score":24,"last_commit_at":51,"category_tags":52,"status":17},1381,"everything-claude-code","affaan-m\u002Feverything-claude-code","everything-claude-code 是一套专为 AI 编程助手（如 Claude Code、Codex、Cursor 等）打造的高性能优化系统。它不仅仅是一组配置文件，而是一个经过长期实战打磨的完整框架，旨在解决 AI 代理在实际开发中面临的效率低下、记忆丢失、安全隐患及缺乏持续学习能力等核心痛点。\n\n通过引入技能模块化、直觉增强、记忆持久化机制以及内置的安全扫描功能，everything-claude-code 能显著提升 AI 在复杂任务中的表现，帮助开发者构建更稳定、更智能的生产级 AI 代理。其独特的“研究优先”开发理念和针对 Token 消耗的优化策略，使得模型响应更快、成本更低，同时有效防御潜在的攻击向量。\n\n这套工具特别适合软件开发者、AI 研究人员以及希望深度定制 AI 工作流的技术团队使用。无论您是在构建大型代码库，还是需要 AI 协助进行安全审计与自动化测试，everything-claude-code 都能提供强大的底层支持。作为一个曾荣获 Anthropic 黑客大奖的开源项目，它融合了多语言支持与丰富的实战钩子（hooks），让 AI 真正成长为懂上",161147,"2026-04-19T23:31:47",[14,13,36],{"id":54,"name":55,"github_repo":56,"description_zh":57,"stars":58,"difficulty_score":59,"last_commit_at":60,"category_tags":61,"status":17},8272,"opencode","anomalyco\u002Fopencode","OpenCode 是一款开源的 AI 编程助手（Coding Agent），旨在像一位智能搭档一样融入您的开发流程。它不仅仅是一个代码补全插件，而是一个能够理解项目上下文、自主规划任务并执行复杂编码操作的智能体。无论是生成全新功能、重构现有代码，还是排查难以定位的 Bug，OpenCode 都能通过自然语言交互高效完成，显著减少开发者在重复性劳动和上下文切换上的时间消耗。\n\n这款工具专为软件开发者、工程师及技术研究人员设计，特别适合希望利用大模型能力来提升编码效率、加速原型开发或处理遗留代码维护的专业人群。其核心亮点在于完全开源的架构，这意味着用户可以审查代码逻辑、自定义行为策略，甚至私有化部署以保障数据安全，彻底打破了传统闭源 AI 助手的“黑盒”限制。\n\n在技术体验上，OpenCode 提供了灵活的终端界面（Terminal UI）和正在测试中的桌面应用程序，支持 macOS、Windows 及 Linux 全平台。它兼容多种包管理工具，安装便捷，并能无缝集成到现有的开发环境中。无论您是追求极致控制权的资深极客，还是渴望提升产出的独立开发者，OpenCode 都提供了一个透明、可信",144296,1,"2026-04-16T14:50:03",[13,27],{"id":63,"github_repo":64,"name":65,"description_en":66,"description_zh":67,"ai_summary_zh":67,"readme_en":68,"readme_zh":69,"quickstart_zh":70,"use_case_zh":71,"hero_image_url":72,"owner_login":73,"owner_name":74,"owner_avatar_url":75,"owner_bio":76,"owner_company":77,"owner_location":78,"owner_email":79,"owner_twitter":77,"owner_website":77,"owner_url":80,"languages":81,"stars":102,"forks":103,"last_commit_at":104,"license":105,"difficulty_score":24,"env_os":106,"env_gpu":107,"env_ram":108,"env_deps":109,"category_tags":115,"github_topics":116,"view_count":24,"oss_zip_url":77,"oss_zip_packed_at":77,"status":17,"created_at":135,"updated_at":136,"faqs":137,"releases":167},9961,"enricoros\u002Fbig-AGI","big-AGI","AI suite powered by state-of-the-art models and providing advanced AI\u002FAGI functions. Includes AI personas, AGI functions, world-class Beam multi-model chats, text-to-image, voice, response streaming, code highlighting and execution, PDF import, presets for developers, much more. Deploy on-prem or in the cloud.","Big-AGI 是一款专为专业人士打造的多模型 AI 工作空间，旨在帮助用户更广阔地思考、更快速地决策并自信地构建项目。它不仅仅是一个聊天机器人，而是一个集成了先进人工智能功能的综合套件，有效解决了单一模型能力有限、复杂任务处理效率低以及本地部署困难等痛点。\n\n无论是需要架构系统的工程师、制定战略的创业者，还是验证假设的研究人员，都能通过 Big-AGI 获得强大的助力。其核心亮点在于独特的\"Beam\"多模型协同对话功能，能够同时调用多个顶级大模型对同一问题进行交叉验证与分析，显著提升回答的准确性与深度。此外，它还支持文生图、语音交互、代码高亮与执行、PDF 文档导入以及丰富的开发者预设模式。\n\nBig-AGI 拥有业界领先的聊天用户体验，既提供便捷的云端服务，也支持通过 Docker 或 Vercel 进行私有化部署，充分保障数据隐私与灵活性。作为一个独立运营、非风险投资驱动的项目，它致力于通过合理的订阅模式反哺开源社区，让免费用户与专业开发者都能享受到前沿的 AGI 技术红利。","\u003Cdiv align=\"center\">\n\n\u003Cimg width=\"256\" height=\"256\" alt=\"Big-AGI Logo\" src=\"https:\u002F\u002Fbig-agi.com\u002Fassets\u002Flogo-bright-github.svg\" \u002F>\n\n\u003Ch1>\u003Ca href=\"https:\u002F\u002Fbig-agi.com\">Big-AGI\u003C\u002Fa>\u003C\u002Fh1>\n\n[![Use Free ⋅ Go Pro](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FUse_Free-Get_Pro-d5ec31?style=for-the-badge&logo=rocket&logoColor=white&labelColor=000)](https:\u002F\u002Fbig-agi.com)\n[![Deploy on Docker](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FSelf--Host-Docker-blue?style=for-the-badge&logo=docker&logoColor=white&labelColor=000)](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fpkgs\u002Fcontainer\u002Fbig-agi)\n[![Deploy on Vercel](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FVercel-Deploy-blue?style=for-the-badge&logo=vercel&logoColor=white&labelColor=000)](https:\u002F\u002Fvercel.com\u002Fnew\u002Fclone?repository-url=https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-agi)\n[![Discord](https:\u002F\u002Fimg.shields.io\u002Fdiscord\u002F1098796266906980422?style=for-the-badge&label=Discord&logo=discord&logoColor=white&labelColor=000000&color=purple)](https:\u002F\u002Fdiscord.gg\u002FMkH4qj2Jp9)\n\u003Cbr\u002F>\n[![GitHub Monthly Commits](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fcommit-activity\u002Fm\u002Fenricoros\u002Fbig-agi?style=for-the-badge&x=3&logo=github&logoColor=white&label=commits&labelColor=000&color=green)](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-agi\u002Fcommits)\n[![GHCR Pulls](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fghcr.io-800k_dl-12b76a?style=for-the-badge&logo=Xdocker&logoColor=white&labelColor=000&color=A8E6CF)](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fpkgs\u002Fcontainer\u002Fbig-agi)\n[![Contributors](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fcontributors\u002Fenricoros\u002Fbig-agi?style=for-the-badge&x=2&logo=Xgithub&logoColor=white&label=cooks&labelColor=000&color=A8E6CF)](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fgraphs\u002Fcontributors)\n[![License: MIT](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FLicense-MIT-A8E6CF?style=for-the-badge&labelColor=000)](https:\u002F\u002Fopensource.org\u002Flicenses\u002FMIT)\n\u003Cbr\u002F>\n\n[![Open an Issue](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FOpen_Issue-AI_Will_Help-ff8c00?style=for-the-badge&logo=fireship&logoColor=fff&labelColor=8b0000)](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-agi\u002Fissues\u002Fnew?template=ai-triage.yml)\n\n[\u002F\u002F]: # ([![Uptime Robot ratio (30 days)](https:\u002F\u002Fimg.shields.io\u002Fuptimerobot\u002Fratio\u002Fm801796948-868b22ed7ceaa0acac4dc765?style=for-the-badge&labelColor=000&color=green)](https:\u002F\u002Fstats.uptimerobot.com\u002F59MXcnmjrM))\n[\u002F\u002F]: # ([![Open Version](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fv\u002Frelease\u002Fenricoros\u002Fbig-AGI?label=Open+Release&style=flat-square&logo=github&logoColor=white&labelColor=000)](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Freleases\u002Flatest))\n[\u002F\u002F]: # (![GitHub Stars](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fstars\u002Fenricoros\u002Fbig-agi?style=flat-square&logo=github&logoColor=white&labelColor=000&color=yellow))\n[\u002F\u002F]: # ([![GitHub Forks](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fforks\u002Fenricoros\u002Fbig-agi?style=flat-square&logo=github&logoColor=white&labelColor=000)](#))\n[\u002F\u002F]: # ([![Follow on X](https:\u002F\u002Fimg.shields.io\u002Ftwitter\u002Ffollow\u002Fenricoros?style=flat-square&logo=X&logoColor=white&labelColor=000&color=000)](https:\u002F\u002Fx.com\u002Fenricoros))\n\n\u003C\u002Fdiv>\n\n\u003Cbr\u002F>\n\n# Big-AGI Open 🧠\n\nThis is the open-source foundation of **Big-AGI**, ___the multi-model AI workspace for experts___.\n\nBig-AGI is the multi-model AI workspace for experts: Engineers architecting systems. Founders making decisions. Researchers validating hypotheses.\nYou need to think broader, decide faster, and build with confidence, then you need Big-AGI.\n\nIt comes packed with **world-class features** like Beam, and is praised for its **best-in-class AI chat UX**.\n**As an independent, non-VC-funded project, Pro subscriptions at $10.99\u002Fmo fund development for everyone, including the free and open-source tiers.**\n\n![LLM Vendors](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F20+_LLM_Services-500+_Models-black?style=for-the-badge&logo=anthropic&logoColor=white&labelColor=purple) \n[![Feature Beam](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FAI--Validation-BEAM-000?style=for-the-badge&labelColor=purple)](https:\u002F\u002Fbig-agi.com\u002Fbeam) \n[![Feature Inspector](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FExpert_Mode-AI_Inspector-000?style=for-the-badge&labelColor=purple)](https:\u002F\u002Fbig-agi.com\u002Finspector)\n\n### What makes Big-AGI different:\n\n**Intelligence**: with [Beam & Merge](https:\u002F\u002Fbig-agi.com\u002Fbeam) for multi-model de-hallucination, native search, and bleeding-edge AI models like Opus 4.7, Nano Banana Pro, Kimi K2.5 or GPT 5.4 -\n**Control**: with personas, data ownership, requests inspection, unlimited usage with API keys, and *no vendor lock-in* -\nand **Speed**: with a local-first, over-powered, zero-latency, madly optimized web app.\n\n\u003Ctable>\n\u003Ctr>\n\u003Ctd align=\"center\" width=\"25%\">\n\u003Cb>🧠 Intelligence\u003C\u002Fb>\u003Cbr\u002F>\n\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FMulti--Model-Trust-4285F4?style=for-the-badge\" alt=\"Multi-Model\"\u002F>\n\u003C\u002Ftd>\n\u003Ctd align=\"center\" width=\"25%\">\n\u003Cb>✨ Experience\u003C\u002Fb>\u003Cbr\u002F>\n\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FClean-UX-34A853?style=for-the-badge\" alt=\"Clean UX\"\u002F>\n\u003C\u002Ftd>\n\u003Ctd align=\"center\" width=\"25%\">\n\u003Cb>⚡ Performance\u003C\u002Fb>\u003Cbr\u002F>\n\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FZero-Latency-EA4335?style=for-the-badge\" alt=\"Zero Latency\"\u002F>\n\u003C\u002Ftd>\n\u003Ctd align=\"center\" width=\"25%\">\n\u003Cb>🔒 Control\u003C\u002Fb>\u003Cbr\u002F>\n\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FNo-Lock--in-FBBC04?style=for-the-badge\" alt=\"No Lock-in\"\u002F>\n\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" valign=\"top\">\nBeam & Merge\u003Cbr\u002F>\nNo context junk\u003Cbr\u002F>\nPurest AI outputs\n\u003C\u002Ftd>\n\u003Ctd align=\"center\" valign=\"top\">\nFlow-state interface\u003Cbr\u002F>\nHighly customizable\u003Cbr\u002F>\nBest-in-class UX\n\u003C\u002Ftd>\n\u003Ctd align=\"center\" valign=\"top\">\nLocal-first\u003Cbr\u002F>\nHighly parallel\u003Cbr\u002F>\nMadly optimized\n\u003C\u002Ftd>\n\u003Ctd align=\"center\" valign=\"top\">\nNo vendor lock-in\u003Cbr\u002F>\nYour API keys\u003Cbr\u002F>\nAI Inspector\n\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003C\u002Ftable>\n\n### Who uses Big-AGI:  \nLoved by engineers, founders, researchers, self-hosters, and IT departments for its power, reliability, and transparency.\n\n\u003Cimg width=\"830\" height=\"370\" alt=\"image\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fenricoros_big-AGI_readme_60702d6cf586.png\" \u002F>\n\nChoose Big-AGI because you don't need another clone or slop - you need an AI tool that scales with you.\n\n### Show me a screenshot:\nSure - here is real-world screeengrab as I'm writing this, while running a Beam to extract SVG from an image with Sonnet 4.5, Opus 4.1, GPT 5.1, Gemini 2.5 Pro, Nano Banana, etc.  \n\u003Cimg alt=\"Real-world screen capture as of Nov 15 2025, 2am\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fenricoros_big-AGI_readme_acb87af9f4b4.png\" \u002F>\n\n\n## Get Started\n\n| Tier                                                 | Best For          | What You Get                                                  | Setup       |\n|------------------------------------------------------|-------------------|---------------------------------------------------------------|-------------|\n| Big-AGI Open (self-host)                             | **IT**            | First to get new models support. Maximum control and privacy. | 5-30 min    |\n| [big-agi.com](https:\u002F\u002Fbig-agi.com) Free              | **Everyone**      | Full core experience, improved Beam, new Personas, best UX.   | **2 min**\\* |\n| **[big-agi.com](https:\u002F\u002Fbig-agi.com) Pro** $10.99\u002Fmo | **Professionals** | Everything + **Sync** across unlimited devices + 1GB storage  | **2 min**\\* |\n\n\\*: **Configuration requires your API keys**. *Big-AGI does not charge for model usage or limit your access*.  \n**Why Pro?** As an independent project, Pro subscriptions fund all development. Early subscribers shape the roadmap directly.    \n\n[![Use Free ⋅ Go Pro](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FUse_Free-Get_Pro-d5ec31?style=for-the-badge&logo=rocket&logoColor=white&labelColor=000)](https:\u002F\u002Fbig-agi.com)\n\n**Self-host and developers** (full control)  \n- Develop locally or self-host with Docker on your own infrastructure – [guide](docs\u002Finstallation.md)  \n- Or fork & run on Vercel:  \n  [![Deploy on Vercel](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FDeploy-black?style=for-the-badge&logo=vercel&logoColor=white&labelColor=000)](https:\u002F\u002Fvercel.com\u002Fnew\u002Fclone?repository-url=https%3A%2F%2Fgithub.com%2Fenricoros%2Fbig-AGI&env=OPENAI_API_KEY&envDescription=Backend%20API%20keys%2C%20optional%20and%20may%20be%20overridden%20by%20the%20UI.&envLink=https%3A%2F%2Fgithub.com%2Fenricoros%2Fbig-AGI%2Fblob%2Fmain%2Fdocs%2Fenvironment-variables.md&project-name=big-AGI)\n\n[\u002F\u002F]: # (**For the latest Big-AGI:**)\n\n[\u002F\u002F]: # (- [**Big-AGI Open**](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Ftree\u002Fmain) - Open Source, latest models and features (main branch))\n\n[\u002F\u002F]: # (- [**Big-AGI Pro**](https:\u002F\u002Fbig-agi.com) - Hosted with Cloud Sync)\n\n---\n\n## Our Philosophy\n\nWe're an independent, non-VC-funded project with a simple belief: **AI should elevate you, not replace you**.\n\nThis is why we built Big-AGI to be **local-first**, madly optimized to 0-latency, launched multi-model first to\ndefeat hallucinations, designed Beam around the **humans in the loop**, re-wrote frameworks and abstractions\nso you **are not vendor locked-in**, and obsessed over a powerful UI that works, just works.\n\nNOTE: this is a powerful tool - if you need a toy UI or clone, this ain't it.\n\n\n---\n\n## Release Notes\n\n👉 **[See the Live Release Notes](https:\u002F\u002Fbig-agi.com\u002Fchanges)**\n- Open 2.0.4: **Hyper Params** **Opus 4.6**, **GPT-5.4**, **Gemini 3.1 Pro**, AWS Bedrock, parameter accuracy, Anthropic continuation\u002FFast mode\n- Open 2.0.3: **Red Carpet** **Kimi K2.5**, **Gemini 3 Flash**, **GPT 5.2**, Google Drive, Inworld, Novita.ai, Speech\u002FUX improvements\n- Open 2.0.2: **Speex** multi-vendor speech synthesis, **Opus 4.5**, **Gemini 3 Pro**, **Nano Banana Pro**, **Grok 4.1**, **GPT-5.1**, **Kimi K2** + 280 fixes\n\n### What's New in 2.0 · Oct 31, 2025 · Open\n\n- **Big-AGI Open** is ready and more productive and faster than ever, with:\n- **Beam 2**: multi-modal, program-based, follow-ups, save presets\n- Top-notch AI models support including **agentic models** and **reasoning models**\n- **Image Generation** and editing with Nano Banana and gpt-image-1\n- **Web Search** with citations for supported models\n- **UI** & Mobile UI overhaul with peeking and side panels\n- And all of the [Big-AGI 2 changes](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F567#issuecomment-2262187617) and more\n- Built for the future, madly optimized\n\n\u003Cimg width=\"830\" height=\"385\" alt=\"image\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fenricoros_big-AGI_readme_947b831c7cd5.png\" \u002F>\n\n#### **Open** links: 👉 [changelog](https:\u002F\u002Fbig-agi.com\u002Fchanges) 👉 [installation](docs\u002Finstallation.md) 👉 [roadmap](https:\u002F\u002Fgithub.com\u002Fusers\u002Fenricoros\u002Fprojects\u002F4\u002Fviews\u002F2) 👉 [documentation](docs\u002FREADME.md)\n\n**For teams and institutions:** Need shared prompts, SSO, or managed deployments? Reach out at enrico@big-agi.com. We're actively collecting requirements from research groups and IT departments.\n\n\u003Cdetails>\n\u003Csummary>5,000 Commits Milestone\u003C\u002Fsummary>\n\nHit 5k commits last week. That's a lot of code.\n\nRecent work has been intense:\n- Chain of thought reasoning across multiple LLMs: **OpenAI o3** and o1, **DeepSeek R1**, **Gemini 2.0 Flash Thinking**, and more\n- Beam is real - ~35% of our users run it daily to compare models\n- New AIX framework lets us scale features we couldn't before\n- UI is faster than ever. Like, terminal-fast\n\nThe new architecture is solid and the speed improvements are real.\n\n![5000e-830px](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fenricoros_big-AGI_readme_d501532b8b53.png)\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>What's New in 1.16.1...1.16.13 · (patch releases)\u003C\u002Fsummary>\n\n- 1.16.13: Docker fix ([#840](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F840))\n- 1.16.12: Dockerfile update ([#840](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F840))\n- 1.16.11: v1 final release, documentation updates\n- 1.16.10: OpenRouter models support\n- 1.16.9: Docker Gemini fix, R1 models support\n- 1.16.8: OpenAI ChatGPT-4o Latest, o1 models support\n- 1.16.7: OpenAI support for GPT-4o 2024-08-06\n- 1.16.6: Groq support for Llama 3.1 models\n- 1.16.5: GPT-4o Mini support\n- 1.16.4: 8192 tokens support for Claude 3.5 Sonnet\n- 1.16.3: Anthropic Claude 3.5 Sonnet model support\n- 1.16.2: Improve web downloads, as text, markdown, or HTML\n- 1.16.2: Proper support for Gemini models\n- 1.16.2: Added the latest Mistral model\n- 1.16.2: Tokenizer support for gpt-4o\n- 1.16.2: Updates to Beam\n- 1.16.1: Support for the new OpenAI GPT-4o 2024-05-13 model\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>What's New in 1.16.0 · May 9, 2024 · Crystal Clear\u003C\u002Fsummary>\n\n- [Beam](https:\u002F\u002Fbig-agi.com\u002Fblog\u002Fbeam-multi-model-ai-reasoning) core and UX improvements based on user feedback\n- Chat cost estimation 💰 (enable it in Labs \u002F hover the token counter)\n- Save\u002Fload chat files with Ctrl+S \u002F Ctrl+O on desktop\n- Major enhancements to the Auto-Diagrams tool\n- YouTube Transcriber Persona for chatting with video content, [#500](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fpull\u002F500)\n- Improved formula rendering (LaTeX), and dark-mode diagrams, [#508](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F508), [#520](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F520)\n- Models update: **Anthropic**, **Groq**, **Ollama**, **OpenAI**, **OpenRouter**, **Perplexity**\n- Code soft-wrap, chat text selection toolbar, 3x faster on Apple silicon, and more [#517](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F517), [507](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fpull\u002F507)\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>3,000 Commits Milestone · April 7, 2024\u003C\u002Fsummary>\n\n![big-AGI Milestone](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fenricoros_big-AGI_readme_9d7403db8b40.png)\n\n- 🥇 Today we \u003Cb>celebrate commit 3000\u003C\u002Fb> in just over one year, and going stronger 🚀\n- 📢️ Thanks everyone for your support and words of love for Big-AGI, we are committed to creating the best AI experiences for everyone.\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>What's New in 1.15.0 · April 1, 2024 · Beam\u003C\u002Fsummary>\n\n- ⚠️ [**Beam**: the multi-model AI chat](https:\u002F\u002Fbig-agi.com\u002Fblog\u002Fbeam-multi-model-ai-reasoning). find better answers, faster - a game-changer for brainstorming, decision-making, and creativity. [#443](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F443)\n- Managed Deployments **Auto-Configuration**: simplify the UI models setup with backend-set models. [#436](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F436)\n- Message **Starring ⭐**: star important messages within chats, to attach them later. [#476](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F476)\n- Enhanced the default Persona\n- Fixes to Gemini models and SVGs, improvements to UI and icons\n- 1.15.1: Support for Gemini Pro 1.5 and OpenAI Turbo models\n- Beast release, over 430 commits, 10,000+ lines changed: [release notes](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Freleases\u002Ftag\u002Fv1.15.0), and changes [v1.14.1...v1.15.0](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fcompare\u002Fv1.14.1...v1.15.0)\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>What's New in 1.14.1 · March 7, 2024 · Modelmorphic\u003C\u002Fsummary>\n\n- **Anthropic** [Claude-3](https:\u002F\u002Fwww.anthropic.com\u002Fnews\u002Fclaude-3-family) model family support. [#443](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F443)\n- New **[Perplexity](https:\u002F\u002Fwww.perplexity.ai\u002F)** and **[Groq](https:\u002F\u002Fgroq.com\u002F)** integration (thanks @Penagwin). [#407](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F407), [#427](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F427)\n- **[LocalAI](https:\u002F\u002Flocalai.io\u002Fmodels\u002F)** deep integration, including support for [model galleries](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F411)\n- **Mistral** Large and Google **Gemini 1.5** support\n- Performance optimizations: runs [much faster](https:\u002F\u002Fx.com\u002Fenricoros\u002Fstatus\u002F1756553038293303434?utm_source=localhost:3000&utm_medium=big-agi), saves lots of power, reduces memory usage\n- Enhanced UX with auto-sizing charts, refined search and folder functionalities, perfected scaling\n- And with more UI improvements, documentation, bug fixes (20 tickets), and developer enhancements\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>What's New in 1.13.0 · Feb 8, 2024 · Multi + Mind\u003C\u002Fsummary>\n\nhttps:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fassets\u002F32999\u002F01732528-730e-41dc-adc7-511385686b13\n\n- **Side-by-Side Split Windows**: multitask with parallel conversations. [#208](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F208)\n- **Multi-Chat Mode**: message everyone, all at once. [#388](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F388)\n- **Export tables as CSV**: big thanks to @aj47. [#392](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fpull\u002F392)\n- Adjustable text size: customize density. [#399](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F399)\n- Dev2 Persona Technology Preview\n- Better looking chats with improved spacing, fonts, and menus\n- More: new video player, [LM Studio tutorial](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fblob\u002Fmain\u002Fdocs\u002Fconfig-local-lmstudio.md) (thanks @aj47), [MongoDB support](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fblob\u002Fmain\u002Fdocs\u002Fdeploy-database.md) (thanks @ranfysvalle02), and speedups\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>What's New in 1.12.0 · Jan 26, 2024 · AGI Hotline\u003C\u002Fsummary>\n\nhttps:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fassets\u002F32999\u002F95ceb03c-945d-4fdd-9a9f-3317beb54f3f\n\n- **Voice Calls**: real-time voice call your personas out of the blue or in relation to a chat [#354](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F354)\n- Support **OpenAI 0125** Models. [#364](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F364)\n- Rename or Auto-Rename chats.  [#222](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F222), [#360](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F360)\n- More control over **Link Sharing** [#356](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F356)\n- **Accessibility** to screen readers [#358](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F358)\n- Export chats to Markdown [#337](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F337)\n- Paste tables from Excel [#286](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F286)\n- Ollama model updates and context window detection fixes [#309](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F309)\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>What's New in 1.11.0 · Jan 16, 2024 · Singularity\u003C\u002Fsummary>\n\nhttps:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fassets\u002F1590910\u002Fa6b8e172-0726-4b03-a5e5-10cfcb110c68\n\n- **Find chats**: search in titles and content, with frequency ranking. [#329](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F329)\n- **Commands**: command auto-completion (type '\u002F'). [#327](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F327)\n- **[Together AI](https:\u002F\u002Fwww.together.ai\u002Fproducts#inference)** inference platform support (good speed and newer models). [#346](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F346)\n- Persona Creator history, deletion, custom creation, fix llm API timeouts\n- Enable adding up to five custom OpenAI-compatible endpoints\n- Developer enhancements: new 'Actiles' framework\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>What's New in 1.10.0 · Jan 6, 2024 · The Year of AGI\u003C\u002Fsummary>\n\n- **New UI**: for both desktop and mobile, sets the stage for future scale. [#201](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F201)\n- **Conversation Folders**: enhanced conversation organization. [#321](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F321)\n- **[LM Studio](https:\u002F\u002Flmstudio.ai\u002F)** support and improved token management\n- Resizable panes in split-screen conversations.\n- Large performance optimizations\n- Developer enhancements: new UI framework, updated documentation for proxy settings on browserless\u002Fdocker\n\n\u003C\u002Fdetails>\n\nFor full details and former releases, check out the [archived versions changelog](docs\u002Fchangelog.md).\n\n## 👉 Supported Models & Integrations\n\nDelightful UX with latest models exclusive features like Beam for **multi-model AI validation**.\n> ![LLM Vendors](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F20_LLM_Services-500+_Models-black?style=for-the-badge&logo=openai&logoColor=white&labelColor=purple) \n> [![Feature Beam](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FAI--Validation-BEAM-000?style=for-the-badge&logo=anthropic&labelColor=purple)](https:\u002F\u002Fbig-agi.com\u002Fbeam)\n\n| ![Advanced AI](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FAdvanced%20AI-32383e?style=for-the-badge&logo=ai&logoColor=white) | ![500+ AI Models](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F500%2B%20AI%20Models-32383e?style=for-the-badge&logo=ai&logoColor=white) | ![Flow-state UX](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FFlow--state%20UX-32383e?style=for-the-badge&logo=flow&logoColor=white) | ![Privacy First](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FPrivacy%20First-32383e?style=for-the-badge&logo=privacy&logoColor=white) | ![Advanced Tools](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FFun%20To%20Use-f22a85?style=for-the-badge&logo=tools&logoColor=white) |  \n|---------------------------------------------------------------------------------------------------------------|-------------------------------------------------------------------------------------------------------------------------|----------------------------------------------------------------------------------------------------------------------|------------------------------------------------------------------------------------------------------------------------|----------------------------------------------------------------------------------------------------------------------| \n| **Chat**\u003Cbr\u002F>**Call**\u003Cbr\u002F>**Beam**\u003Cbr\u002F>**Draw**, ...                                                          | Local & Cloud\u003Cbr\u002F>Open & Closed\u003Cbr\u002F>Cheap & Heavy\u003Cbr\u002F>Google, Mistral, ...                                              | Attachments\u003Cbr\u002F>Diagrams\u003Cbr\u002F>Multi-Chat\u003Cbr\u002F>Mobile-first UI                                                          | Stored Locally\u003Cbr\u002F>Easy self-Host\u003Cbr\u002F>Local actions\u003Cbr\u002F>Data = Gold                                                    | AI Personas\u003Cbr\u002F>Voice Modes\u003Cbr\u002F>Screen Capture\u003Cbr\u002F>Camera + OCR                                                      |\n\n![big-AGI screenshot](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fenricoros_big-AGI_readme_3f5efa3ecf86.png)\n\n### AI Models & Vendors\n\nConfigure 100s of AI models from 20+ providers:\n\n| **AI models**       | _supported vendors_                                                                                                                                                                                                                                                                                                                                                                                                                                                                        |\n|:--------------------|:-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|\n| Opensource Servers  | [LocalAI](https:\u002F\u002Flocalai.io\u002F) · [Ollama](https:\u002F\u002Follama.com\u002F)                                                                                                                                                                                                                                                                                                                                                                                                                             |\n| Local Servers       | [LM Studio](https:\u002F\u002Flmstudio.ai\u002F) (non-open)                                                                                                                                                                                                                                                                                                                                                                                                                                               |\n| Multimodal services | [Anthropic](https:\u002F\u002Fanthropic.com) · [AWS Bedrock](https:\u002F\u002Faws.amazon.com\u002Fbedrock\u002F) · [Azure](https:\u002F\u002Fazure.microsoft.com\u002Fen-us\u002Fproducts\u002Fai-services\u002Fopenai-service) · [Google Gemini](https:\u002F\u002Fai.google.dev\u002F) · [OpenAI](https:\u002F\u002Fplatform.openai.com\u002Fdocs\u002Foverview)                                                                                                                                                                                                                       |\n| LLM services        | [Alibaba](https:\u002F\u002Fwww.alibabacloud.com\u002Fen\u002Fproduct\u002Fmodelstudio) · [DeepSeek](https:\u002F\u002Fdeepseek.com) · [Groq](https:\u002F\u002Fwow.groq.com\u002F) · [Mistral](https:\u002F\u002Fmistral.ai\u002F) · [Moonshot](https:\u002F\u002Fwww.moonshot.cn\u002F) · [OpenPipe](https:\u002F\u002Fopenpipe.ai\u002F) · [OpenRouter](https:\u002F\u002Fopenrouter.ai\u002F) · [Perplexity](https:\u002F\u002Fwww.perplexity.ai\u002F) · [Together AI](https:\u002F\u002Fwww.together.ai\u002F) · [xAI](https:\u002F\u002Fx.ai\u002F) · [Z.ai](https:\u002F\u002Fz.ai\u002F) |\n| OpenAI-compatible   | Any OpenAI-compatible endpoint - models, pricing, and capabilities are auto-detected                                                                                                                                                                                                                                                                                                                                                                                                         |\n| Image services      | OpenAI · Google Gemini (Nano Banana) · LocalAI                                                                                                                                                                                                                                                                                                                                                                                                                                              |\n| Speech services     | [ElevenLabs](https:\u002F\u002Felevenlabs.io) · [Inworld](https:\u002F\u002Finworld.ai) · [OpenAI TTS](https:\u002F\u002Fplatform.openai.com\u002Fdocs\u002Fguides\u002Ftext-to-speech) · LocalAI · Browser (Web Speech API)                                                                                                                                                                                                                                                                                                            |\n\n### Additional Integrations\n\n| **More**      | _integrations_                                                                                                 |\n|:--------------|:---------------------------------------------------------------------------------------------------------------| \n| Web Browse    | [Browserless](https:\u002F\u002Fwww.browserless.io\u002F) · [Puppeteer](https:\u002F\u002Fpptr.dev\u002F)-based                              |\n| Web Search    | [Google CSE](https:\u002F\u002Fprogrammablesearchengine.google.com\u002F)                                                     |\n| Code Editors  | [CodePen](https:\u002F\u002Fcodepen.io\u002Fpen\u002F) · [StackBlitz](https:\u002F\u002Fstackblitz.com\u002F) · [JSFiddle](https:\u002F\u002Fjsfiddle.net\u002F) |\n| Observability | [Helicone](https:\u002F\u002Fwww.helicone.ai)                                                                            |\n\n---\n\n## 🚀 Installation\n\nSelf-host with Docker, deploy on Vercel, or develop locally. Full setup guide:\n\n[![Installation Guide](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FInstallation%20Guide-blue?style=for-the-badge&logo=read-the-docs&logoColor=white)](docs\u002Finstallation.md)\n\nOr use the hosted version at [big-agi.com](https:\u002F\u002Fbig-agi.com) with your API keys.\n\n---\n\n## 👋 Community & Contributing\n\n### Connect\n\n[![Official Discord](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fenricoros_big-AGI_readme_81b3c76106f3.png)](https:\u002F\u002Fdiscord.gg\u002FMkH4qj2Jp9)\n\n⭐ [Star the repo](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-agi) if Big-AGI is useful to you\n\n### Contribute\n\n**🤖 AI-Powered Issue Assistance**\n\nWhen you open an issue, our custom AI triage system (powered by [Claude Code](https:\u002F\u002Fgithub.com\u002Fanthropics\u002Fclaude-code-action) with Big-AGI architecture documentation) analyzes it, searches the codebase, and provides solutions - typically within 30 minutes. We've trained the system on our modules and subsystems so it handles most issues effectively. Your feedback drives development!\n\n[![Open an Issue](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FOpen_Issue-AI_Will_Help-ff8c00?style=for-the-badge&logo=fireship&logoColor=fff&labelColor=8b0000)](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-agi\u002Fissues\u002Fnew?template=ai-triage.yml)\n[![Request Feature](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FRequest_Feature-Roadmap_Idea-orange?style=for-the-badge&logo=lightbulb&logoColor=white)](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-agi\u002Fissues\u002Fnew?&template=roadmap-request.md)\n\n[![Good First Issues](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FGood_First_Issues-Start-blue?style=for-the-badge&logo=github&logoColor=white)](https:\u002F\u002Fgithub.com\u002Fusers\u002Fenricoros\u002Fprojects\u002F4\u002Fviews\u002F4)\n[![Customization](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FFork_&_Customize-Your_Own-purple?style=for-the-badge&logo=git&logoColor=white)](docs\u002Fcustomizations.md)\n[![Roadmap](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FOpen_Roadmap-View-0366d6?style=for-the-badge&logo=github&logoColor=white)](https:\u002F\u002Fgithub.com\u002Fusers\u002Fenricoros\u002Fprojects\u002F4\u002Fviews\u002F2)\n\n#### Contributors\n\n\u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-agi\u002Fgraphs\u002Fcontributors\">\n  \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fenricoros_big-AGI_readme_ccf29c9cb70d.png\" \u002F>\n\u003C\u002Fa>\n\n---\n\n## License\n\nMIT License · [Third-Party Notices](src\u002Fmodules\u002F3rdparty\u002FTHIRD_PARTY_NOTICES.md)\n\n**2023-2026** · [Enrico Ros](https:\u002F\u002Fwww.enricoros.com) × [Token Fabrics](https:\u002F\u002Fwww.tokenfabrics.com)\n","\u003Cdiv align=\"center\">\n\n\u003Cimg width=\"256\" height=\"256\" alt=\"Big-AGI Logo\" src=\"https:\u002F\u002Fbig-agi.com\u002Fassets\u002Flogo-bright-github.svg\" \u002F>\n\n\u003Ch1>\u003Ca href=\"https:\u002F\u002Fbig-agi.com\">Big-AGI\u003C\u002Fa>\u003C\u002Fh1>\n\n[![免费使用 ⋅ 升级专业版](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F免费使用-升级专业版-d5ec31?style=for-the-badge&logo=rocket&logoColor=white&labelColor=000)](https:\u002F\u002Fbig-agi.com)\n[![在 Docker 上部署](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F自托管-Docker-blue?style=for-the-badge&logo=docker&logoColor=white&labelColor=000)](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fpkgs\u002Fcontainer\u002Fbig-agi)\n[![在 Vercel 上部署](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FVercel-部署-blue?style=for-the-badge&logo=vercel&logoColor=white&labelColor=000)](https:\u002F\u002Fvercel.com\u002Fnew\u002Fclone?repository-url=https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-agi)\n[![Discord](https:\u002F\u002Fimg.shields.io\u002Fdiscord\u002F1098796266906980422?style=for-the-badge&label=Discord&logo=discord&logoColor=white&labelColor=000000&color=purple)](https:\u002F\u002Fdiscord.gg\u002FMkH4qj2Jp9)\n\u003Cbr\u002F>\n[![GitHub 每月提交量](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fcommit-activity\u002Fm\u002Fenricoros\u002Fbig-agi?style=for-the-badge&x=3&logo=github&logoColor=white&label=提交&labelColor=000&color=green)](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-agi\u002Fcommits)\n[![GHCR 下载量](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fghcr.io-80万次下载-12b76a?style=for-the-badge&logo=Xdocker&logoColor=white&labelColor=000&color=A8E6CF)](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fpkgs\u002Fcontainer\u002Fbig-agi)\n[![贡献者](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fcontributors\u002Fenricoros\u002Fbig-agi?style=for-the-badge&x=2&logo=Xgithub&logoColor=white&label=厨师&labelColor=000&color=A8E6CF)](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fgraphs\u002Fcontributors)\n[![许可证：MIT](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F许可证-MIT-A8E6CF?style=for-the-badge&labelColor=000)](https:\u002F\u002Fopensource.org\u002Flicenses\u002FMIT)\n\u003Cbr\u002F>\n\n[![提交问题](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F提交问题-AI会帮忙-ff8c00?style=for-the-badge&logo=fireship&logoColor=fff&labelColor=8b0000)](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-agi\u002Fissues\u002Fnew?template=ai-triage.yml)\n\n[\u002F\u002F]: # ([![Uptime Robot 比例（30天）](https:\u002F\u002Fimg.shields.io\u002Fuptimerobot\u002Fratio\u002Fm801796948-868b22ed7ceaa0acac4dc765?style=for-the-badge&labelColor=000&color=green)](https:\u002F\u002Fstats.uptimerobot.com\u002F59MXcnmjrM))\n[\u002F\u002F]: # ([![开放版本](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fv\u002Frelease\u002Fenricoros\u002Fbig-AGI?label=开放+发布&style=flat-square&logo=github&logoColor=white&labelColor=000)](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Freleases\u002Flatest))\n[\u002F\u002F]: # (![GitHub 星标](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fstars\u002Fenricoros\u002Fbig-agi?style=flat-square&logo=github&logoColor=white&labelColor=000&color=yellow))\n[\u002F\u002F]: # ([![GitHub 分支数](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fforks\u002Fenricoros\u002Fbig-agi?style=flat-square&logo=github&logoColor=white&labelColor=000)](#))\n[\u002F\u002F]: # ([![关注 X](https:\u002F\u002Fimg.shields.io\u002Ftwitter\u002Ffollow\u002Fenricoros?style=flat-square&logo=X&logoColor=white&labelColor=000&color=000)](https:\u002F\u002Fx.com\u002Fenricoros))\n\n\u003C\u002Fdiv>\n\n\u003Cbr\u002F>\n\n# Big-AGI 开源 🧠\n\n这是 **Big-AGI** 的开源基础，___专为专家打造的多模型 AI 工作空间___。\n\nBig-AGI 是专为专家设计的多模型 AI 工作空间：工程师构建系统、创始人做出决策、研究人员验证假设。无论你需要更广阔的视野、更快的决策速度，还是充满信心地进行开发，Big-AGI 都能满足你的需求。\n\n它内置了像 Beam 这样的**世界级功能**，并以其**一流的 AI 聊天用户体验**而备受赞誉。\n**作为独立且未获风投支持的项目，每月 10.99 美元的专业订阅将用于资助所有用户的开发，包括免费和开源版本。**\n\n![LLM 供应商](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F20+_LLM_服务-500+_模型-black?style=for-the-badge&logo=anthropic&logoColor=white&labelColor=purple) \n[![Beam 功能](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FAI--验证-BEAM-000?style=for-the-badge&labelColor=purple)](https:\u002F\u002Fbig-agi.com\u002Fbeam) \n[![Inspector 功能](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F专家模式-AI_Inspector-000?style=for-the-badge&labelColor=purple)](https:\u002F\u002Fbig-agi.com\u002Finspector)\n\n### Big-AGI 的独特之处：\n\n**智能性**：通过 [Beam & Merge](https:\u002F\u002Fbig-agi.com\u002Fbeam) 实现多模型去幻觉、原生搜索，以及 Opus 4.7、Nano Banana Pro、Kimi K2.5 或 GPT 5.4 等尖端 AI 模型——\n**可控性**：提供个性化角色、数据所有权、请求审查、无限使用 API 密钥，并且*无厂商锁定*——\n以及**速度**：采用本地优先、性能超强、零延迟、高度优化的 Web 应用程序。\n\n\u003Ctable>\n\u003Ctr>\n\u003Ctd align=\"center\" width=\"25%\">\n\u003Cb>🧠 智能性\u003C\u002Fb>\u003Cbr\u002F>\n\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F多模型-信任-4285F4?style=for-the-badge\" alt=\"多模型\"\u002F>\n\u003C\u002Ftd>\n\u003Ctd align=\"center\" width=\"25%\">\n\u003Cb>✨ 体验\u003C\u002Fb>\u003Cbr\u002F>\n\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F简洁-UX-34A853?style=for-the-badge\" alt=\"简洁 UX\"\u002F>\n\u003C\u002Ftd>\n\u003Ctd align=\"center\" width=\"25%\">\n\u003Cb>⚡ 性能\u003C\u002Fb>\u003Cbr\u002F>\n\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F零延迟-EA4335?style=for-the-badge\" alt=\"零延迟\"\u002F>\n\u003C\u002Ftd>\n\u003Ctd align=\"center\" width=\"25%\">\n\u003Cb>🔒 控制\u003C\u002Fb>\u003Cbr\u002F>\n\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F无锁-定-FBBC04?style=for-the-badge\" alt=\"无锁-定\"\u002F>\n\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" valign=\"top\">\nBeam & Merge\u003Cbr\u002F>\n无冗余上下文\u003Cbr\u002F>\n纯净的 AI 输出\n\u003C\u002Ftd>\n\u003Ctd align=\"center\" valign=\"top\">\n心流界面\u003Cbr\u002F>\n高度可定制\u003Cbr\u002F>\n一流 UX\n\u003C\u002Ftd>\n\u003Ctd align=\"center\" valign=\"top\">\n本地优先\u003Cbr\u002F>\n高度并行\u003Cbr\u002F>\n极致优化\n\u003C\u002Ftd>\n\u003Ctd align=\"center\" valign=\"top\">\n无厂商锁定\u003Cbr\u002F>\n您的 API 密钥\u003Cbr\u002F>\nAI Inspector\n\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003C\u002Ftable>\n\n### 谁在使用 Big-AGI：\n工程师、创始人、研究人员、自托管用户以及 IT 部门都因其强大、可靠和透明的特点而青睐 Big-AGI。\n\n\u003Cimg width=\"830\" height=\"370\" alt=\"图片\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fenricoros_big-AGI_readme_60702d6cf586.png\" \u002F>\n\n选择 Big-AGI，因为您不需要另一个克隆或低质量的产品——您需要的是能够与您共同成长的 AI 工具。\n\n### 给我看看截图吧：\n当然——这是我正在撰写本文时的实际屏幕截图，当时我正在运行 Beam，从一张图片中提取 SVG 文件，使用的模型包括 Sonnet 4.5、Opus 4.1、GPT 5.1、Gemini 2.5 Pro、Nano Banana 等。\n\u003Cimg alt=\"2025年11月15日凌晨2点的真实屏幕截图\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fenricoros_big-AGI_readme_acb87af9f4b4.png\" \u002F>\n\n## 开始使用\n\n| 等级                                                 | 适合人群          | 包含内容                                                  | 设置时间       |\n|------------------------------------------------------|-------------------|---------------------------------------------------------------|-------------|\n| Big-AGI Open（自托管）                             | **IT 专业人士**    | 最先获得新模型支持。最高级别的控制与隐私保护。               | 5-30 分钟    |\n| [big-agi.com](https:\u002F\u002Fbig-agi.com) 免费              | **所有人**        | 完整的核心体验、优化的 Beam 功能、全新 Persona、最佳用户体验。   | **2 分钟**\\* |\n| **[big-agi.com](https:\u002F\u002Fbig-agi.com) Pro** $10.99\u002F月 | **专业人士**      | 包含所有功能 + **跨设备同步** + 1GB 存储空间                | **2 分钟**\\* |\n\n\\*: **配置需要您的 API 密钥**。*Big-AGI 不会对模型使用收费，也不会限制您的访问权限*。  \n**为什么选择 Pro？** 作为一个独立项目，Pro 订阅收入将用于支持所有开发工作。早期订阅者可以直接影响产品路线图。\n\n[![免费使用 ⋅ 升级 Pro](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FUse_Free-Get_Pro-d5ec31?style=for-the-badge&logo=rocket&logoColor=white&labelColor=000)](https:\u002F\u002Fbig-agi.com)\n\n**自托管与开发者**（完全掌控）  \n- 您可以在本地开发或通过 Docker 在自己的基础设施上进行自托管 – [指南](docs\u002Finstallation.md)  \n- 或者在 Vercel 上 fork 并运行：  \n  [![在 Vercel 部署](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FDeploy-black?style=for-the-badge&logo=vercel&logoColor=white&labelColor=000)](https:\u002F\u002Fvercel.com\u002Fnew\u002Fclone?repository-url=https%3A%2F%2Fgithub.com%2Fenricoros%2Fbig-AGI&env=OPENAI_API_KEY&envDescription=Backend%20API%20keys%2C%20optional%20and%20may%20be%20overridden%20by%20the%20UI.&envLink=https%3A%2F%2Fgithub.com%2Fenricoros%2Fbig-AGI%2Fblob%2Fmain%2Fdocs%2Fenvironment-variables.md&project-name=big-AGI)\n\n[\u002F\u002F]: # (**获取最新版 Big-AGI：**)\n\n[\u002F\u002F]: # (- [**Big-AGI Open**]（https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Ftree\u002Fmain） - 开源，包含最新模型与功能（主分支）)\n\n[\u002F\u002F]: # (- [**Big-AGI Pro**]（https:\u002F\u002Fbig-agi.com） - 托管服务，支持云端同步）\n\n---\n\n## 我们的理念\n\n我们是一个独立的、未获风险投资支持的项目，秉持着一个简单的信念：**AI 应该提升你，而不是取代你**。\n\n因此，我们打造了 Big-AGI，使其成为“本地优先”的系统，极致优化以实现零延迟；我们率先推出多模态模型来对抗幻觉问题；围绕“人类参与”设计了 Beam 功能；重新编写框架与抽象层，确保您不会被厂商锁定；并专注于构建一个强大且真正好用的用户界面。\n\n注意：这是一个强大的工具——如果您只是想要一个玩具般的界面或克隆版本，那这里不适合您。\n\n\n---\n\n## 发布说明\n\n👉 **[查看实时发布说明](https:\u002F\u002Fbig-agi.com\u002Fchanges)**\n- Open 2.0.4：**超参数** **Opus 4.6**、**GPT-5.4**、**Gemini 3.1 Pro**、AWS Bedrock、参数精度、Anthropic 继续对话\u002F快速模式\n- Open 2.0.3：**红毯版** **Kimi K2.5**、**Gemini 3 Flash**、**GPT 5.2**、Google Drive、Inworld、Novita.ai、语音及用户体验改进\n- Open 2.0.2：**Speex** 多供应商语音合成、**Opus 4.5**、**Gemini 3 Pro**、**Nano Banana Pro**、**Grok 4.1**、**GPT-5.1**、**Kimi K2** + 280 处修复\n\n### 2.0 版本更新 · 2025年10月31日 · Open\n\n- **Big-AGI Open** 已经准备就绪，比以往更加高效、快速，具备以下特性：\n- **Beam 2**：多模态、基于程序的交互、后续追问、保存预设\n- 支持顶尖 AI 模型，包括**代理型模型**和**推理型模型**\n- **图像生成**与编辑功能，采用 Nano Banana 和 gpt-image-1\n- **网络搜索**功能，并为支持的模型提供引用来源\n- **UI** 及移动端 UI 全面升级，新增预览与侧边栏功能\n- 以及所有 [Big-AGI 2 的变更](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F567#issuecomment-2262187617) 和更多\n- 专为未来而建，经过极致优化\n\n\u003Cimg width=\"830\" height=\"385\" alt=\"image\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fenricoros_big-AGI_readme_947b831c7cd5.png\" \u002F>\n\n#### **Open** 相关链接：👉 [变更日志](https:\u002F\u002Fbig-agi.com\u002Fchanges) 👉 [安装指南](docs\u002Finstallation.md) 👉 [路线图](https:\u002F\u002Fgithub.com\u002Fusers\u002Fenricoros\u002Fprojects\u002F4\u002Fviews\u002F2) 👉 [文档](docs\u002FREADME.md)\n\n**面向团队与机构：** 需要共享提示词、SSO 或管理式部署吗？请联系我们 enrico@big-agi.com。我们正在积极收集来自研究团队和 IT 部门的需求。\n\n\u003Cdetails>\n\u003Csummary>5,000 次提交里程碑\u003C\u002Fsummary>\n\n上周达到了 5,000 次提交。这可是一大堆代码呢。\n\n最近的工作非常密集：\n- 跨多个 LLM 的思维链推理：**OpenAI o3** 和 o1、**DeepSeek R1**、**Gemini 2.0 Flash Thinking** 等\n- Beam 已经落地——约 35% 的用户每天使用它来比较不同模型\n- 新的 AIX 框架使我们能够扩展以前无法实现的功能\n- UI 比以往任何时候都更快，快到像终端一样迅速\n\n新的架构非常稳固，速度提升也确实显著。\n\n![5000e-830px](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fenricoros_big-AGI_readme_d501532b8b53.png)\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>1.16.1 至 1.16.13 的更新...（补丁版本）\u003C\u002Fsummary>\n\n- 1.16.13：Docker 修复（[#840](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F840)）\n- 1.16.12：Dockerfile 更新（[#840](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F840)）\n- 1.16.11：v1 正式发布，文档更新\n- 1.16.10：支持 OpenRouter 模型\n- 1.16.9：修复 Gemini 的 Docker 问题，支持 R1 模型\n- 1.16.8：支持 OpenAI ChatGPT-4o Latest 和 o1 模型\n- 1.16.7：支持 OpenAI GPT-4o，发布日期为 2024 年 8 月 6 日\n- 1.16.6：支持 Groq 对 Llama 3.1 模型的支持\n- 1.16.5：支持 GPT-4o Mini\n- 1.16.4：支持 Claude 3.5 Sonnet 的 8192 个 token\n- 1.16.3：支持 Anthropic Claude 3.5 Sonnet 模型\n- 1.16.2：改进网页下载功能，可导出为文本、Markdown 或 HTML 格式\n- 1.16.2：对 Gemini 模型提供更完善的支持\n- 1.16.2：添加了最新的 Mistral 模型\n- 1.16.2：支持 gpt-4o 的分词器\n- 1.16.2：对 Beam 进行了多项更新\n- 1.16.1：支持全新的 OpenAI GPT-4o，发布日期为 2024 年 5 月 13 日\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>1.16.0 版本更新 · 2024 年 5 月 9 日 · 清晰明了\u003C\u002Fsummary>\n\n- 基于用户反馈对 [Beam](https:\u002F\u002Fbig-agi.com\u002Fblog\u002Fbeam-multi-model-ai-reasoning) 的核心功能和用户体验进行了改进\n- 聊天成本估算 💰（可在 Labs 中启用，将鼠标悬停在 token 计数器上）\n- 桌面端可通过 Ctrl+S \u002F Ctrl+O 保存\u002F加载聊天记录\n- 自动图表工具得到重大增强\n- YouTube Transcriber Persona，可用于与视频内容对话，[#500](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fpull\u002F500)\n- 改进了公式渲染（LaTeX）以及暗色模式下的图表，[#508](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F508)、[#520](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F520)\n- 模型更新：**Anthropic**、**Groq**、**Ollama**、**OpenAI**、**OpenRouter**、**Perplexity**\n- 代码自动换行、聊天文本选择工具栏、在 Apple Silicon 上速度提升 3 倍等 [#517](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F517)、[507](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fpull\u002F507)\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>3,000 次提交里程碑 · 2024 年 4 月 7 日\u003C\u002Fsummary>\n\n![big-AGI 里程碑](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fenricoros_big-AGI_readme_9d7403db8b40.png)\n\n- 🥇 今天我们在短短一年多时间内迎来了第3000次提交，并且势头更加强劲 🚀\n- 📢 感谢大家对Big-AGI的支持与鼓励，我们致力于为所有人打造最佳的AI体验。\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>1.15.0 版本更新 · 2024年4月1日 · Beam\u003C\u002Fsummary>\n\n- ⚠️ [**Beam**：多模型AI聊天](https:\u002F\u002Fbig-agi.com\u002Fblog\u002Fbeam-multi-model-ai-reasoning)。更快找到更优答案——这将彻底改变头脑风暴、决策制定和创意生成的方式。[#443](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F443)\n- 托管部署 **自动配置**：通过后端预设模型简化UI模型设置。[#436](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F436)\n- 消息 **星标⭐**：在聊天中为重要消息添加星标，方便后续快速查找。[#476](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F476)\n- 默认人格形象优化\n- 修复了Gemini模型和SVG显示问题，改进了UI及图标\n- 1.15.1版本：支持Gemini Pro 1.5和OpenAI Turbo模型\n- Beast版本，超过430次提交，修改代码超过1万行：[发布说明](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Freleases\u002Ftag\u002Fv1.15.0)，以及[v1.14.1...v1.15.0](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fcompare\u002Fv1.14.1...v1.15.0)的变更记录\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>1.14.1 版本更新 · 2024年3月7日 · Modelmorphic\u003C\u002Fsummary>\n\n- 支持 **Anthropic** [Claude-3](https:\u002F\u002Fwww.anthropic.com\u002Fnews\u002Fclaude-3-family) 系列模型。[#443](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F443)\n- 新增 **[Perplexity](https:\u002F\u002Fwww.perplexity.ai\u002F)** 和 **[Groq](https:\u002F\u002Fgroq.com\u002F)** 集成（感谢@Penagwin）。[#407](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F407), [#427](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F427)\n- **[LocalAI](https:\u002F\u002Flocalai.io\u002Fmodels\u002F)** 深度集成，包括对[模型库](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F411)的支持\n- 支持 **Mistral Large** 和谷歌 **Gemini 1.5**\n- 性能优化：运行速度显著提升[见推文](https:\u002F\u002Fx.com\u002Fenricoros\u002Fstatus\u002F1756553038293303434?utm_source=localhost:3000&utm_medium=big-agi)，大幅节省电力并降低内存占用\n- 用户体验增强：图表自动调整大小，搜索与文件夹功能更加完善，缩放效果更佳\n- 此外还有更多UI改进、文档更新、20个问题修复以及开发者工具增强\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>1.13.0 版本更新 · 2024年2月8日 · Multi + Mind\u003C\u002Fsummary>\n\nhttps:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fassets\u002F32999\u002F01732528-730e-41dc-adc7-511385686b13\n\n- **并排分割窗口**：同时进行多项任务，展开多条对话。[#208](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F208)\n- **多聊模式**：一次向所有人发送消息。[#388](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F388)\n- **表格导出为CSV**：特别感谢@aj47。[#392](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fpull\u002F392)\n- 文字大小可调：自定义内容密度。[#399](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F399)\n- Dev2人格技术预览\n- 聊天界面焕然一新，间距、字体和菜单均得到优化\n- 其他新增功能：全新视频播放器、[LM Studio教程](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fblob\u002Fmain\u002Fdocs\u002Fconfig-local-lmstudio.md)（感谢@aj47）、[MongoDB支持](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fblob\u002Fmain\u002Fdocs\u002Fdeploy-database.md)（感谢@ranfysvalle02），以及性能加速\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>1.12.0 版本更新 · 2024年1月26日 · AGI热线\u003C\u002Fsummary>\n\nhttps:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fassets\u002F32999\u002F95ceb03c-945d-4fdd-9a9f-3317beb54f3f\n\n- **语音通话**：随时随地实时语音呼叫你的人格角色，无论是单独通话还是与聊天关联。[#354](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F354)\n- 支持 **OpenAI 0125** 模型。[#364](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F364)\n- 重命名或自动重命名聊天。[#222](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F222), [#360](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F360)\n- 更好地控制 **链接分享** [#356](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F356)\n- 增强 **无障碍访问**，支持屏幕阅读器 [#358](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F358)\n- 聊天内容导出为Markdown [#337](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F337)\n- 可从Excel粘贴表格 [#286](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F286)\n- Ollama模型更新及上下文窗口检测修复 [#309](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F309)\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>1.11.0 版本更新 · 2024年1月16日 · 奇点\u003C\u002Fsummary>\n\nhttps:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fassets\u002F1590910\u002Fa6b8e172-0726-4b03-a5e5-10cfcb110c68\n\n- **查找聊天**：按标题和内容搜索，并按频率排序。[#329](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F329)\n- **命令**：输入“\u002F”即可实现命令自动补全。[#327](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F327)\n- 支持 **[Together AI](https:\u002F\u002Fwww.together.ai\u002Fproducts#inference)** 推理平台（速度快，模型较新）。[#346](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F346)\n- 人格创建历史、删除功能、自定义创建，修复LLM API超时问题\n- 允许添加最多五个兼容OpenAI的自定义API端点\n- 开发者工具增强：推出全新‘Actiles’框架\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>1.10.0 版本更新 · 2024年1月6日 · AGI之年\u003C\u002Fsummary>\n\n- **全新UI**：桌面端与移动端统一设计，为未来扩展奠定基础。[#201](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F201)\n- **对话文件夹**：进一步优化对话组织方式。[#321](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F321)\n- 支持 **[LM Studio](https:\u002F\u002Flmstudio.ai\u002F)** 并改进令牌管理\n- 分屏对话中的窗格可调整大小。\n- 大规模性能优化\n- 开发者工具增强：推出全新UI框架，更新无浏览器\u002FDocker环境下的代理设置文档\n\n\u003C\u002Fdetails>\n\n如需了解完整详情及过往版本信息，请查阅[归档版本变更日志](docs\u002Fchangelog.md)。\n\n## 👉 支持的模型与集成\n\n采用最新模型的愉悦用户体验，独家功能如 Beam，用于**多模型 AI 验证**。\n> ![LLM 供应商](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F20_LLM_Services-500+_Models-black?style=for-the-badge&logo=openai&logoColor=white&labelColor=purple) \n> [![功能：Beam](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FAI--Validation-BEAM-000?style=for-the-badge&logo=anthropic&labelColor=purple)](https:\u002F\u002Fbig-agi.com\u002Fbeam)\n\n| ![先进 AI](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FAdvanced%20AI-32383e?style=for-the-badge&logo=ai&logoColor=white) | ![500+ AI 模型](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F500%2B%20AI%20Models-32383e?style=for-the-badge&logo=ai&logoColor=white) | ![流畅状态 UX](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FFlow--state%20UX-32383e?style=for-the-badge&logo=flow&logoColor=white) | ![隐私优先](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FPrivacy%20First-32383e?style=for-the-badge&logo=privacy&logoColor=white) | ![高级工具](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FFun%20To%20Use-f22a85?style=for-the-badge&logo=tools&logoColor=white) |  \n|---------------------------------------------------------------------------------------------------------------|-------------------------------------------------------------------------------------------------------------------------|----------------------------------------------------------------------------------------------------------------------|------------------------------------------------------------------------------------------------------------------------|----------------------------------------------------------------------------------------------------------------------| \n| **聊天**\u003Cbr\u002F>**通话**\u003Cbr\u002F>**Beam**\u003Cbr\u002F>**绘图**, ...                                                          | 本地与云端\u003Cbr\u002F>开放与封闭\u003Cbr\u002F>轻量与重型\u003Cbr\u002F>Google、Mistral 等                                              | 附件\u003Cbr\u002F>图表\u003Cbr\u002F>多聊模式\u003Cbr\u002F>移动端优先界面                                                          | 数据本地存储\u003Cbr\u002F>轻松自托管\u003Cbr\u002F>本地操作\u003Cbr\u002F>数据即财富                                                    | AI 角色\u003Cbr\u002F>语音模式\u003Cbr\u002F>屏幕截图\u003Cbr\u002F>摄像头 + OCR                                                      |\n\n![big-AGI 截图](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fenricoros_big-AGI_readme_3f5efa3ecf86.png)\n\n### AI 模型与供应商\n\n配置来自 20 多家提供商的数百种 AI 模型：\n\n| **AI 模型**       | _支持的供应商_                                                                                                                                                                                                                                                                                                                                                                                                                                                                        |\n|:--------------------|:-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|\n| 开源服务器  | [LocalAI](https:\u002F\u002Flocalai.io\u002F) · [Ollama](https:\u002F\u002Follama.com\u002F)                                                                                                                                                                                                                                                                                                                                                                                                                             |\n| 本地服务器       | [LM Studio](https:\u002F\u002Flmstudio.ai\u002F)（非开源）                                                                                                                                                                                                                                                                                                                                                                                                                                               |\n| 多模态服务       | [Anthropic](https:\u002F\u002Fanthropic.com) · [AWS Bedrock](https:\u002F\u002Faws.amazon.com\u002Fbedrock\u002F) · [Azure](https:\u002F\u002Fazure.microsoft.com\u002Fen-us\u002Fproducts\u002Fai-services\u002Fopenai-service) · [Google Gemini](https:\u002F\u002Fai.google.dev\u002F) · [OpenAI](https:\u002F\u002Fplatform.openai.com\u002Fdocs\u002Foverview)                                                                                                                                                                                                                       |\n| LLM 服务        | [Alibaba](https:\u002F\u002Fwww.alibabacloud.com\u002Fen\u002Fproduct\u002Fmodelstudio) · [DeepSeek](https:\u002F\u002Fdeepseek.com) · [Groq](https:\u002F\u002Fwow.groq.com\u002F) · [Mistral](https:\u002F\u002Fmistral.ai\u002F) · [Moonshot](https:\u002F\u002Fwww.moonshot.cn\u002F) · [OpenPipe](https:\u002F\u002Fopenpipe.ai\u002F) · [OpenRouter](https:\u002F\u002Fopenrouter.ai\u002F) · [Perplexity](https:\u002F\u002Fwww.perplexity.ai\u002F) · [Together AI](https:\u002F\u002Fwww.together.ai\u002F) · [xAI](https:\u002F\u002Fx.ai\u002F) · [Z.ai](https:\u002F\u002Fz.ai\u002F) |\n| OpenAI 兼容      | 任何 OpenAI 兼容的端点——模型、定价和功能都会自动检测                                                                                                                                                                                                                                                                                                                                                                                                         |\n| 图像服务      | OpenAI · Google Gemini（Nano Banana）· LocalAI                                                                                                                                                                                                                                                                                                                                                                                                                                              |\n| 语音服务     | [ElevenLabs](https:\u002F\u002Felevenlabs.io) · [Inworld](https:\u002F\u002Finworld.ai) · [OpenAI TTS](https:\u002F\u002Fplatform.openai.com\u002Fdocs\u002Fguides\u002Ftext-to-speech) · LocalAI · 浏览器（Web Speech API）                                                                                                                                                                                                                                                                                                            |\n\n### 其他集成\n\n| **更多**      | _集成_                                                                                                 |\n|:--------------|:---------------------------------------------------------------------------------------------------------------| \n| 网页浏览    | [Browserless](https:\u002F\u002Fwww.browserless.io\u002F) · 基于 [Puppeteer](https:\u002F\u002Fpptr.dev\u002F)                              |\n| 网络搜索    | [Google 自定义搜索](https:\u002F\u002Fprogrammablesearchengine.google.com\u002F)                                                     |\n| 代码编辑器  | [CodePen](https:\u002F\u002Fcodepen.io\u002Fpen\u002F) · [StackBlitz](https:\u002F\u002Fstackblitz.com\u002F) · [JSFiddle](https:\u002F\u002Fjsfiddle.net\u002F) |\n| 可观测性    | [Helicone](https:\u002F\u002Fwww.helicone.ai)                                                                            |\n\n---\n\n## 🚀 安装\n\n使用 Docker 自行部署、在 Vercel 上部署，或在本地开发。完整安装指南：\n\n[![安装指南](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F安装%20指南-blue?style=for-the-badge&logo=read-the-docs&logoColor=white)](docs\u002Finstallation.md)\n\n或者使用托管版本 [big-agi.com](https:\u002F\u002Fbig-agi.com)，只需提供您的 API 密钥即可。\n\n---\n\n## 👋 社区与贡献\n\n### 联系我们\n\n[![官方 Discord](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fenricoros_big-AGI_readme_81b3c76106f3.png)](https:\u002F\u002Fdiscord.gg\u002FMkH4qj2Jp9)\n\n⭐ 如果 Big-AGI 对您有帮助，请为该项目加星 ([Star the repo](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-agi))\n\n### 参与贡献\n\n**🤖 AI 驱动的问题协助**\n\n当您提交问题时，我们的自定义 AI 分类系统（基于 [Claude Code](https:\u002F\u002Fgithub.com\u002Fanthropics\u002Fclaude-code-action)，并结合 Big-AGI 架构文档）会对其进行分析、搜索代码库，并通常在 30 分钟内提供解决方案。我们已针对项目中的模块和子系统对该系统进行了训练，因此它能够高效处理大多数问题。您的反馈将推动项目的持续发展！\n\n[![提交问题](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FOpen_Issue-AI_Will_Help-ff8c00?style=for-the-badge&logo=fireship&logoColor=fff&labelColor=8b0000)](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-agi\u002Fissues\u002Fnew?template=ai-triage.yml)\n[![请求功能](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FRequest_Feature-Roadmap_Idea-orange?style=for-the-badge&logo=lightbulb&logoColor=white)](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-agi\u002Fissues\u002Fnew?&template=roadmap-request.md)\n\n[![适合初学者的问题](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FGood_First_Issues-Start-blue?style=for-the-badge&logo=github&logoColor=white)](https:\u002F\u002Fgithub.com\u002Fusers\u002Fenricoros\u002Fprojects\u002F4\u002Fviews\u002F4)\n[![自定义](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FFork_&_Customize-Your_Own-purple?style=for-the-badge&logo=git&logoColor=white)](docs\u002Fcustomizations.md)\n[![路线图](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FOpen_Roadmap-View-0366d6?style=for-the-badge&logo=github&logoColor=white)](https:\u002F\u002Fgithub.com\u002Fusers\u002Fenricoros\u002Fprojects\u002F4\u002Fviews\u002F2)\n\n#### 贡献者\n\n\u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-agi\u002Fgraphs\u002Fcontributors\">\n  \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fenricoros_big-AGI_readme_ccf29c9cb70d.png\" \u002F>\n\u003C\u002Fa>\n\n---\n\n## 许可证\n\nMIT 许可证 · [第三方声明](src\u002Fmodules\u002F3rdparty\u002FTHIRD_PARTY_NOTICES.md)\n\n**2023–2026** · [Enrico Ros](https:\u002F\u002Fwww.enricoros.com) × [Token Fabrics](https:\u002F\u002Fwww.tokenfabrics.com)","# Big-AGI 快速上手指南\n\nBig-AGI 是一款专为专家打造的多模型 AI 工作区，支持本地优先部署、多模型并行推理（Beam）、零延迟交互及完全的数据控制权。本指南将帮助开发者快速完成自托管部署。\n\n## 环境准备\n\n在开始之前，请确保您的系统满足以下要求：\n\n*   **操作系统**：Linux, macOS, 或 Windows (需安装 WSL2 或 Docker Desktop)\n*   **核心依赖**：\n    *   **Docker** & **Docker Compose** (推荐方式，版本建议 v20.10+)\n    *   或者 **Node.js** (v18.18+ 及 npm\u002Fyarn\u002Fpnpm) 用于源码开发\n*   **API 密钥**：准备至少一个 LLM 服务商的 API Key（如 OpenAI, Anthropic, Google Gemini, Azure 等）。Big-AGI 本身不提供免费模型，需用户自行配置密钥。\n*   **网络环境**：由于需要拉取 Docker 镜像及连接海外 AI 服务商，建议配置好网络代理或使用国内加速源。\n\n## 安装步骤\n\n推荐使用 **Docker** 进行部署，这是最稳定且维护成本最低的方式。\n\n### 方法一：使用 Docker Compose（推荐）\n\n1.  **创建项目目录并下载配置文件**\n    在终端执行以下命令创建目录并获取官方 `docker-compose.yml` 文件：\n    ```bash\n    mkdir big-agi && cd big-agi\n    curl -O https:\u002F\u002Fraw.githubusercontent.com\u002Fenricoros\u002Fbig-AGI\u002Fmain\u002Fdocker-compose.yml\n    ```\n    *(注：如果下载速度慢，可手动访问 GitHub 仓库复制内容保存为 `docker-compose.yml`)*\n\n2.  **配置环境变量**\n    创建 `.env` 文件以存储您的 API 密钥和服务配置：\n    ```bash\n    cp .env.example .env\n    ```\n    编辑 `.env` 文件，填入您的 API Key：\n    ```bash\n    # 示例：配置 OpenAI Key\n    OPENAI_API_KEY=sk-your-actual-api-key-here\n    \n    # 可选：配置其他服务商\n    ANTHROPIC_API_KEY=sk-ant-...\n    GOOGLE_GENERATIVE_AI_API_KEY=...\n    \n    # 可选：自定义端口 (默认为 3000)\n    # NEXT_PUBLIC_APP_PORT=3000\n    ```\n\n3.  **启动服务**\n    在包含 `docker-compose.yml` 和 `.env` 的目录下运行：\n    ```bash\n    docker compose up -d\n    ```\n    等待镜像拉取完成（首次运行可能需要几分钟）。\n\n4.  **验证安装**\n    查看容器状态：\n    ```bash\n    docker compose ps\n    ```\n    确保状态显示为 `Up`。\n\n### 方法二：使用 Vercel 一键部署（适合云端试用）\n\n如果您希望快速在云端体验而无需管理服务器，可以点击下方的按钮直接部署到 Vercel：\n\n[![Deploy on Vercel](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FDeploy-black?style=for-the-badge&logo=vercel&logoColor=white&labelColor=000)](https:\u002F\u002Fvercel.com\u002Fnew\u002Fclone?repository-url=https%3A%2F%2Fgithub.com%2Fenricoros%2Fbig-AGI&env=OPENAI_API_KEY&envDescription=Backend%20API%20keys%2C%20optional%20and%20may%20be%20overridden%20by%20the%20UI.&envLink=https%3A%2F%2Fgithub.com%2Fenricoros%2Fbig-AGI%2Fblob%2Fmain%2Fdocs%2Fenvironment-variables.md&project-name=big-AGI)\n\n*注意：Vercel 免费版可能有函数执行时长限制，生产环境建议使用 Docker 自托管。*\n\n## 基本使用\n\n部署完成后，即可开始使用 Big-AGI 的核心功能。\n\n1.  **访问界面**\n    打开浏览器访问：`http:\u002F\u002Flocalhost:3000` (如果您修改了端口，请使用对应端口)。\n\n2.  **配置模型提供商**\n    *   进入界面后，点击左侧边栏的 **Settings** (设置图标)。\n    *   在 **Models** 选项卡中，您会看到已自动读取 `.env` 中的密钥。\n    *   如果没有自动读取，可手动粘贴 API Key。\n    *   勾选您想要使用的模型（支持同时勾选多个模型，如 GPT-4, Claude 3.5, Gemini Pro 等）。\n\n3.  **开始对话与 Beam 模式**\n    *   **普通对话**：在输入框输入问题，选择单个模型进行聊天。\n    *   **Beam 模式 (核心特色)**：\n        1.  点击输入框上方的 **Beam** 图标（或多模型选择器）。\n        2.  选择 2 个或更多不同的模型（例如同时选择 `GPT-4o` 和 `Claude 3.5 Sonnet`）。\n        3.  发送消息。系统将并行调用所有选中模型，并在同一界面展示它们的回答，方便您对比结果、消除幻觉并获取最佳方案。\n\n4.  **高级功能**\n    *   **人物设定 (Personas)**：在设置中加载预设的专家角色（如“高级架构师”、“代码审查员”）。\n    *   **本地搜索**：启用 Web Search 插件（需在设置中配置搜索引擎 API），让模型具备联网检索能力。\n    *   **数据导出**：所有对话记录均存储在本地浏览器或您的数据库中，支持导出为 Markdown、PDF 或 JSON。\n\n现在您已经成功部署并使用了 Big-AGI。作为本地优先的工具，您的所有数据都将保留在您自己的控制之下。","一位全栈工程师正在紧急重构一个遗留的金融数据处理模块，需要同时完成代码迁移、逻辑验证和技术文档更新。\n\n### 没有 big-AGI 时\n- **模型切换繁琐**：为了兼顾代码生成的准确性和逻辑推理的深度，开发者必须在多个网页标签页间频繁切换不同的 AI 服务商，上下文无法互通。\n- **验证成本高昂**：生成的复杂算法代码缺乏即时验证机制，只能手动复制到本地 IDE 运行测试，一旦报错需反复往返修改，耗时极长。\n- **文档处理割裂**：原有的几百页 PDF 技术规范无法直接作为参考投喂给模型，只能人工摘录关键段落，极易遗漏重要约束条件。\n- **协作视角单一**：在架构决策时缺乏“多模型辩论”机制，仅凭单一模型的输出做决定，容易陷入思维盲区，增加系统风险。\n\n### 使用 big-AGI 后\n- **一站式多模型协作**：利用 Beam 功能同时调用多个顶尖模型进行“背对背”代码生成与逻辑互验，在一个界面内即可综合最优解，无需切换窗口。\n- **代码即时执行高亮**：直接在对话框中高亮显示生成的代码并一键执行测试，实时捕获运行时错误并自动修复，将调试循环从分钟级缩短至秒级。\n- **原生 PDF 深度导入**：直接将完整的 PDF 技术规范拖入对话，big-AGI 自动解析全文内容作为上下文约束，确保重构代码严格符合原有业务逻辑。\n- **专家模式辅助决策**：启用预设的“架构师”人格与 AI 检查器（Inspector），让不同模型针对同一方案进行对抗性分析，快速识别潜在的性能瓶颈与安全漏洞。\n\nbig-AGI 通过整合多模型协同、本地代码执行及深度文档解析能力，将原本碎片化的开发流程转化为高效、可信的闭环工作流。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fenricoros_big-AGI_3f5efa3e.png","enricoros","Enrico Ros","https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002Fenricoros_8a1713ce.jpg","Token magician ✨ Fuse, melt, forge, ship, and build the world",null,"California","enrico.ros@gmail.com","https:\u002F\u002Fgithub.com\u002Fenricoros",[82,86,90,94,98],{"name":83,"color":84,"percentage":85},"TypeScript","#3178c6",81,{"name":87,"color":88,"percentage":89},"JavaScript","#f1e05a",18.2,{"name":91,"color":92,"percentage":93},"CSS","#663399",0.7,{"name":95,"color":96,"percentage":97},"Shell","#89e051",0.1,{"name":99,"color":100,"percentage":101},"Dockerfile","#384d54",0,6924,1570,"2026-04-19T03:51:19","MIT","Linux, macOS, Windows","未说明 (作为 Web 应用运行，计算负载主要在云端 API 或用户自有的本地基础设施上)","未说明",{"notes":110,"python":111,"dependencies":112},"Big-AGI 是一个本地优先的 Web 应用程序，主要通过 Docker 自托管或部署在 Vercel 上运行。它本身不包含大型模型权重，而是作为界面连接外部 LLM 服务（如 OpenAI, Anthropic, Google 等），因此需要用户自行提供 API Key。硬件需求主要取决于运行 Docker 容器的基础服务器性能，而非特定的 GPU 型号或显存大小。","未说明 (通过 Docker 或 Vercel 部署，环境由容器或平台管理)",[113,114],"Docker","Node.js (隐含，基于 Next.js\u002FVercel 架构)",[36,27,13],[117,118,119,120,121,122,123,124,125,126,127,128,129,130,131,132,133,134],"agi","gpt","ai-agents","ai-suite","ai-workspace","anthropic-api","deepseek-api","gemini-api","gpt-5","multi-modal","multi-model","nano-banana","openai-responses-api","perplexity-api","sonnet-4-5","xai-api","librechat","openwebui","2026-03-27T02:49:30.150509","2026-04-20T12:55:22.706516",[138,143,148,152,157,162],{"id":139,"question_zh":140,"answer_zh":141,"source_url":142},44730,"如何启用和使用自定义角色（Personas）功能？","自定义角色功能已对网站用户开放，包括免费层级（Free tier）。该功能支持保存自定义系统提示词，并具备同步功能。您可以直接尝试使用。此外，使用代码 BIG_HALLOWEEN_25 可获得一个月的 Pro 版免费试用。","https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F35",{"id":144,"question_zh":145,"answer_zh":146,"source_url":147},44731,"项目中是否支持网页搜索（Web Search）功能？","是的，目前搜索功能仅在特定的模式下可用。具体来说，它仅在 `\u002FReAct` 模式或通过 `\u002Fbrowse` 命令时生效。目前该功能尚未作为通用工具连接到 GPT-4 的直接调用中，但在 ReAct 代理循环中可以正常工作。","https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F98",{"id":149,"question_zh":150,"answer_zh":151,"source_url":147},44732,"配置 Google 网页搜索时遇到错误，需要检查哪些设置？","确保您已正确切换模型版本（如尝试切换到 3.5），并在设置中填入了有效的 Google API Key 和 Google CSE ID。在测试期间，建议将 API Key 权限设置为“无限制（unrestricted）”，并在 Google Cloud 控制台中启用相关的额外 API 服务。如果仍然无效，请确认该功能是否在当前选定的模型模式下被支持（通常需在 ReAct 模式下使用）。",{"id":153,"question_zh":154,"answer_zh":155,"source_url":156},44733,"使用 Anthropic 模型时遇到 \"Skill execution error: unavailable\" 或 \"pause_turn\" 错误怎么办？","这通常是由于 Anthropic 服务器在使用大量工具（如 web_search, web_fetch）时返回的中间状态（pause_turn）或伪影导致的。维护者已修复了相关验证逻辑，新版本的应用程序现在会静默忽略可恢复的 `code_execution_tool_result_error` 错误，而不再将其显示为阻断性报错。请更新到最新版本以解决此问题。","https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F1010",{"id":158,"question_zh":159,"answer_zh":160,"source_url":161},44734,"Azure OpenAI 新添加的模型（如 GPT-5, o3 Pro）报错 \"Resource Not Found\" 如何解决？","该问题是由于应用程序错误地构建了请求 URL，将过时的硬编码路径和旧版 `api-version` 附加到了用户配置的端点后面。维护者已经合并了修复补丁， streamlined 了环境变量配置并修正了 URL 构建逻辑。请拉取最新的 GitHub 代码或更新部署版本，即可支持 GPT-5 及更高版本的 Azure OpenAI 模型。","https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F828",{"id":163,"question_zh":164,"answer_zh":165,"source_url":166},44735,"是否支持 GPT-4 Vision 及多模态输入功能？","是的，GPT-4 Vision 支持已完全实现。系统采用了多部分（multipart）架构和 `oai-chat-vision` 接口。在 UI 层面，图片作为附加附件显示在输入框（Composer）下方。如果选择的模型不支持图像输入，该功能会自动禁用。移动端也做了相应的适配（如可展开列表或横向滚动）。","https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F215",[168,173,178,183,188,193,198,203,208,213,218,223,228,233,238,243,248,253,258,263],{"id":169,"version":170,"summary_zh":171,"released_at":172},352187,"v2.0.4","# Big-AGI 2.0.4 - 超参数\n\n### 精确度至关重要。\n\n本次发布专注于优化模型配置：在不同提供商以及通过 OpenRouter 和 AWS 等第三方平台时，确保推理力度、详细程度和温度等参数范围的准确性。此外，还加入了最新的前沿模型、全新的云平台，并进一步深化了对 Anthropic 的支持。\n\n在此特别感谢 @Slava-256，正是他让 Big-AGI 在此次版本中实现了业界领先的模型支持。无论是提交详细的议题以确保推理力度范围的准确性（[#940](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F940)、[#944](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F944)），还是提升详细程度的准确性（[#947](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F947)、[#948](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F948)），亦或是完善温度参数的支持（[#953](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F953)），乃至推动对 Opus 4.6 的支持（[#962](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F962)），以及 [更多工作（共关闭了 31 个议题！）](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues?q=state%3Aclosed%20author%3ASlava-256）。Slava 细致入微的测试、精准的报告、深入的发现以及参考代码，正是促使 2.0.4 版本能够专注于精确参数支持的关键所在。\n\n\u003Cimg width=\"1792\" height=\"1024\" alt=\"0 - A2 - 高调白色场景，洁净无瑕，采用英雄式构图；一只水豚的特写镜头，栩栩如生地由水晶玻璃雕琢而成；水豚姿态优雅，佩戴时尚黑色太阳镜，稳稳地摆出姿势\" src=\"https:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002F257df06f-d365-4bf6-a81a-a9e74f97c0fe\" \u002F>\n\n### 模型与参数\n- Anthropic **Opus 4.6** 具有自适应思维能力及 100 万 token 上下文长度，**Sonnet 4.6** 及其新特性\n- Google **Gemini 3.1 Pro**、**Nano Banana 2**\n- OpenAI **GPT-5.4** 系列，支持 100 万 token 上下文及推理控制\n- xAI **Grok 4.20**\n- 新增对 **Z.ai** GLM 模型的支持，并可进行思维模式的配置\n- **API 验证过的参数**：推理力度、预算、详细程度、温度、工具等参数均已在实际 API 上进行了测试\n- **模型特定的控制项**：每个模型仅暴露其真正支持的参数值\n- **自动化扫描测试**贯穿原生 API，确保参数定义始终保持最新状态\n\n### 精通 Anthropic API\n- 可配置的网页搜索与抓取深度，以及动态搜索功能\n- API 控制选项，例如 **基于美国的推理** 和 **Opus 的快速模式**（速度提升 2.5 倍，但成本高达 6 倍）\n- 支持长输出的续写功能（暂停回合支持）\n\n### AWS Bedrock 支持（新增）\n- 可通过 IAM 凭证或单个 AWS Bedrock 令牌进行身份验证\n- 原生支持三种 AWS Bedrock 协议：Anthropic 模型通过 Invoke 接口调用，Amazon Nova 模型通过 Converse 接口调用，其他模型则通过兼容 OpenAI 的 ChatCompletions 接口调用\n\n### 其他新增功能\n- 任何消息均可附加文件，提供无损图像选项（请注意：在 Gemini 模型上可能会导致文件体积非常大）\n- 富文本复制功能、推理轨迹控制\n- 系统主题模式跟随功能\n- [移动焦点模式](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F894)\n\n## 新 Contr","2026-03-25T02:21:17",{"id":174,"version":175,"summary_zh":176,"released_at":177},352188,"v2.0.3","# Big-AGI 2.0.3 - 红毯版\n\n## 新功能\n\n### **模型更新**\n本次发布着重提升模型精度：优化参数配置以匹配各模型供应商，新增原生 xAI 风格的响应 API，引入全新模型供应商，并支持所有最新模型，其中包括已成经典的 Kimi 2.5，特别适合代码相关任务。\n- **Kimi K2.5**：具备图像理解与推理能力\n- **Gemini 3 Flash**：支持四级思维模式（高、中、低、极低）\n- **GPT 5.2**：超高推理能力，原生集成搜索功能\n- **GPT Image 1.5**：用于图像生成\n- **Novita.ai**：全面支持该提供商，并实现自动能力检测\n- **LMStudio**：支持模型能力展示及图像处理\n- **xAI Responses**：原生协议支持，集成 X 后端搜索与代码执行功能\n\n### **其他新特性**\n- **Google Drive** 集成：可上传文档、表格、图片，并实现最优的 LLM 转换\n- **Inworld** 语音合成，支持流式播放\n- 模型复制功能，便于自定义配置\n- AI 注入器、可折叠模型分组、PDF 自动 OCR\n- “按显示内容复制”功能，以及清理模式下的消息重新排序\n- Anthropic 和 OpenAI API 修复、安全补丁，以及对 CJK\u002FIME 输入的支持\n\n![203-a](https:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002F07e959ed-8610-44f6-ba58-c3d18e24b0aa)\n\n**完整变更日志**：https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fcompare\u002Fv2.0.2...v2.0.3\n\n## 开始使用\n现已在 [big-agi.com](https:\u002F\u002Fbig-agi.com) 上线，也可通过 Docker 获取，或从源码自行部署。","2026-02-02T23:56:42",{"id":179,"version":180,"summary_zh":181,"released_at":182},352189,"v2.0.2","# Big-AGI 2.0.2 - Speex\n\n## 新功能\n\n### **Speex：多供应商语音合成**\n对文本转语音功能进行了全面重构，支持多家服务商：\n- **Web Speech API** —— 无需配置的浏览器原生语音（离线可用）\n- **LocalAI** —— 自托管 TTS，支持 Kokoro、Bark 和 Piper 后端\n- **ElevenLabs** —— 高质量语音合成与克隆\n- **OpenAI TTS** —— gpt-4o-mini-tts 支持语音指令，另有 tts-1 和 tts-1-hd\n\n引擎会根据您配置的 LLM 服务自动检测，无需重复输入 API 密钥。基于优先级的回退机制确保语音功能始终可用。\n\n### **其他新增内容**\n- 完全支持 Claude Opus 4.5\n- 星标模型以便快速访问\n- 持续提升稳定性\n\n\u003Cimg width=\"1792\" height=\"1024\" alt=\"release-cover-v2 0 0\" src=\"https:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002F35ea95cf-896e-403f-bfe4-3aca59fd11a1\" \u002F>\n\n**完整更新日志**：https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fcompare\u002Fv2.0.1...v2.0.2\n\n## 开始使用\n现已在 [big-agi.com](https:\u002F\u002Fbig-agi.com) 上架，也可通过 Docker 获取，或从源代码自行部署。","2025-12-01T11:54:48",{"id":184,"version":185,"summary_zh":186,"released_at":187},352190,"v2.0.1","# Big-AGI 2.0.1 - 重磅更新\n\n## 新功能概览\n\n### **最新模型**\n- 支持代码执行的 **Gemini 3 Pro**\n- 用于可控图像生成的 **Nano Banana Pro**\n- **Grok 4.1 Fast**、**GPT-5.1** 系列以及 **Kimi K2 Thinking**\n- 更新了超过17家服务商的所有模型定义\n\n### **增强特性**\n- **图像生成** 现在除了 OpenAI，还支持 Azure 和 LocalAI\n- **OpenRouter** 可自动检测模型能力与推理方式\n- 调用记录会自动保存\n- 支持生成 **角色形象图**\n- Beam 中新增搜索功能\n\n### **优化改进**\n- 错误恢复能力显著提升（连接问题减少85%）\n- 共有280次提交包含修复与优化\n- 引入 [AI 自动化功能开发](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-agi\u002Fissues\u002Fnew?template=ai-triage.yml)，便于社区贡献\n\n![release-cover-v2 0 0](https:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002F35ea95cf-896e-403f-bfe4-3aca59fd11a1)\n\n**完整变更日志**: https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fcompare\u002Fv2.0.0...v2.0.1\n\n## 开始使用\n即刻访问 [big-agi.com](https:\u002F\u002Fbig-agi.com) 获取，或通过 Docker 部署，亦可从源码自行部署。","2025-11-24T12:18:54",{"id":189,"version":190,"summary_zh":191,"released_at":192},352191,"v2.0.0","# 🚀 Big-AGI 2.0.0 - 开放、免费与专业版\n\n### 历时一年打造，专为专业人士重新设计。\n\n**Big-AGI 2.0** 现已发布。这不仅仅是一次更新；我们重构了核心组件，大幅优化了用户界面，并正式推出三档订阅模式，以支持项目的下一阶段发展。\n\n它专为以下人群打造：**工程师**用于系统架构设计、**创始人**用于决策制定、**研究人员**用于验证假设，以及 IT 部门用于将最高价值的技术部署到企业中。经过 **数千小时的工程开发** 后，我们终于交付了大家期待的产品：一个以本地优先、供应商中立为核心理念，并针对零延迟进行了极致优化的专业级 AI 工作空间。\n\n\u003Cimg width=\"1792\" height=\"1024\" alt=\"release-cover-v2 0 0\" src=\"https:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002F35ea95cf-896e-403f-bfe4-3aca59fd11a1\" \u002F>\n\n## 新功能亮点\n\n### **Beam 2：多模态智能**\n可同时跨多个模型进行程序化推理，支持 **图像**、**后续对话** 和 **保存的预设**。\n\n### **专业级功能**\n- **推理模型**：Anthropic Claude Opus 4.1、Gemini 2.5 Pro、OpenAI GPT-5\u002Fo3\u002Fo1、xAI Grok 4 等。\n- **原生网页搜索**，并为支持的模型提供引用来源。\n- **图像生成**：支持 Nano Banana 和 gpt-image-1，还包含图像编辑功能。\n- **本地文件系统集成**：只需拖放文件即可添加上下文、对比差异并自动保存回原位置。\n- 全面重制桌面端和移动端用户界面。\n\n### **云同步（仅限 Pro 版）**\n历时 10 个月独立开发完成。可在无限设备间实现实时同步，提供 1GB 存储空间。数据完全由您掌控。\n\n### **技术改进**\n- **本地优先架构重设计**：尽可能在您的浏览器中运行所有功能，实现零延迟，避免厂商锁定。\n- **多部分\u002F多模态处理**：几乎每个子系统都经过重新设计和大幅优化。\n- 更多改进——详情请参阅 [big-agi.com\u002Fchanges](https:\u002F\u002Fbig-agi.com\u002Fchanges)。\n\n### 三档订阅方案\n- **Big-AGI Open**（自托管）：率先体验新模型，拥有最高控制权，深受 IT 部门青睐。\n- **Big-AGI Free**（big-agi.com）：完整的核心功能，包括 Beam、Personas 及最佳用户体验。\n- **Big-AGI Pro**（每月 $10.99 起）：涵盖所有功能，并新增 **云同步** 功能，支持无限设备及 1GB 存储空间。\n\n作为一家独立、未获风投资助的项目，Pro 订阅收入将用于支持所有开发工作，包括免费和开源版本的持续演进。\n\n\u003Ca href=\"https:\u002F\u002Fbig-agi.com\">\n  \u003Cimg width=\"210\" height=\"68\" alt=\"image\" src=\"https:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002Fb2f8a7b8-415f-4c92-b228-4f5a54fe2bdd\" \u002F>\n\u003C\u002Fa>  \n\n### 初始启动截图\n\u003Cimg width=\"2116\" height=\"1274\" alt=\"image\" src=\"https:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002Fdcd3811d-8993-4827-8e25-f7c47929dbb9\" \u002F>\n\n\n## 新贡献者\n欢迎并感谢我们的 12 位新贡献者，他们首次为 Big-AGI 做出了贡献！你们的代码如今已成为全球专业人士每日信赖的一部分。\n\n* @powyncify 在 https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fpull 中完成了首次贡献（UI 改进）。","2025-10-28T03:11:22",{"id":194,"version":195,"summary_zh":196,"released_at":197},352192,"v1.16.0","# 🚀 big-AGI 1.16.0 - 晶莹剔透\n\n### 在与 AI 的交互中实现清晰透明。\n\n**big-AGI 1.16.0** —— **晶莹剔透**—— 带来了多项增强功能和新特性，旨在让每一次与 big-AGI 的互动都更加顺畅、直观，也更出色——一如既往。我们希望您会喜欢全面更新的模型，以及全新的成本跟踪功能（请刷新模型以加载最新价格，现已支持 Anthropic、OpenAI 和 OpenRouter），这将帮助您节省开支，因为凡是可衡量的事物，都可以进行优化。\n\n![release-cover-v1.16.0](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fassets\u002F32999\u002F02472393-96d5-47b8-96f4-15d16b193c14)\n\n## 变更内容\n\n- **聊天成本估算**：为支持的模型网络显示成本估算*；鼠标悬停可查看 token 信息。\n![feature-turn-cost-2](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fassets\u002F32999\u002Fa33c5190-a670-47d1-ba7b-93a0bd78b2cb)\n- 根据您的反馈对 [Beam](https:\u002F\u002Fbig-agi.com\u002Fblog\u002Fbeam-multi-model-ai-reasoning) 进行了优化：自动选择并记忆模型网络，简化了用户体验 #470\n- 对 **Auto-Diagrams 工具** 进行了重大改进\n- 使用 Ctrl+S 和 Ctrl+O 实现 **保存\u002F加载聊天文件** #466\n- @joriskalz：新增 **YouTube 转录器** 角色：可转录 YouTube 视频并与之对话（请在角色选择器中将其显示出来）#500\n- @mludvig：通过 Docker 多架构构建，使 **Apple Silicon 平台上的运行速度提升至三倍** #507\n- 更多改进：完美公式\u002FLaTeX 渲染 #508，代码软换行 #517，聊天文本选择工具栏，深色模式修复 #520\n- 文档修正由 @dandv 和 @PrivTEC 完成，并体现在 #521、#522 和 #525 中\n- 更新了 Anthropic*、Groq、Ollama、OpenAI*、OpenRouter* 和 Perplexity 的模型\n- 针对开发者：更新了 LLM 数据结构\n- [完整变更日志](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fcompare\u002Fv1.15.1...v1.16.0)\n\n## 新贡献者\n* 热烈欢迎并感谢 @mludvig、@PrivTEC 和 @dandv 为 big-AGI 做出的首次贡献！\n\n## 总结\n通过提升透明度和用户体验，改善您的生活品质。我们很高兴能够整合社区贡献，包括全新的 YouTube 转录器角色，以及更快的多架构（ARM、x86）构建版本。\n\n立即升级，体验更加清晰、快速、直观的 big-AGI。您可以通过源码、Docker 或在 [big-agi.com](https:\u002F\u002Fbig-agi.com) 上的免费实例获取。🚀","2024-05-09T06:26:30",{"id":199,"version":200,"summary_zh":201,"released_at":202},352193,"v1.15.1","## 小版本更新\n- 新增对刚发布的 Google 1.5 Pro 模型的支持\n- 新增对刚发布的 OpenAI 2024-04-09 模型的支持\n- 改进了桌面版聊天的保存与加载功能（Ctrl + S、Ctrl + O）\n- 其他：新增 docs\u002Finstallation.md 文件，进行了一些改进\n\n## 新贡献者\n* @lunamidori5 在 https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fpull\u002F484 中完成了首次贡献\n* @dogmatic69 在 https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fpull\u002F497 中完成了首次贡献\n\n**完整变更日志**：https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fcompare\u002Fv1.15.0...v1.15.1","2024-04-10T07:54:57",{"id":204,"version":205,"summary_zh":206,"released_at":207},352194,"v1.15.0","# 🚀 big-AGI 1.15.0 - Beam\n\n### 推出多模型 AI 聊天功能。\n\n我们非常高兴地宣布 **big-AGI 1.15.0 - Beam**。[Beam 是 big-AGI 中一种全新的聊天模式](https:\u002F\u002Fbig-agi.com\u002Fblog\u002Fbeam-multi-model-ai-reasoning)，它通过整合多个模型的集体智慧，帮助用户找到更优质的答案。这对于头脑风暴、决策制定和创意激发来说，无疑是一次颠覆性的变革。\n\n如今，“突破性”这个词常常被滥用，但这次推出的新技术确实名副其实——而且这还是专属于各位 big-AGI 爱好者的全球独家功能！在超过 400 次提交和逾 1 万行代码改动的支持下，big-AGI 1.15 为 AI 聊天树立了一个新的里程碑。\n\n![release-cover-v1 15 0 r](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fassets\u002F32999\u002F421ebefa-d0ce-4818-a380-ee3f03bf7d82)\n\n- **Beam**：开启多模型 AI 聊天体验。#470\n- **托管部署自动配置**：通过为托管部署提供自动配置功能，简化您的设置流程——只需在后端设置环境变量，前端便会自动加载所有模型；本地模型同样适用，例如 [LocalAI](https:\u002F\u002Flocalai.io)。#436\n- **消息加星⭐与筛选**：您可以为重要消息添加星标，未来还将支持聊天会话的标记与筛选功能。目前该功能已实现，但我们仍在测试相关的 UI 交互模式。#476\n- **角色形象优化**：默认的 AI 角色形象进一步提升。\n- **UI 与模型修复**：修复了 Gemini 模型及 SVG 图像相关的问题，并对 UI 和图标进行了优化，以带来更加流畅的使用体验。\n\n### 针对开发者：\n- **命令式 LLM 模型发现**：已从 React 方式迁移至基于响应式钩子的命令式代码，从而支持 #436 功能。\n- 完整变更日志：https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fcompare\u002Fv1.14.1...v1.15.0\n\n## 总结\n\n1.15.0 - Beam 是迄今为止最具影响力的 big-AGI 版本。借助多模型的强大能力，Beam 让用户能够探索多元视角、综合生成最优回复，并不断突破 AI 辅助头脑风暴、决策制定和创意激发的边界。\n\n立即体验 AI 聊天的未来——升级至 Beam 吧！您可以通过源码、Docker 或直接访问 [big-agi.com](https:\u002F\u002Fbig-agi.com) 获取。加入我们在 [Discord](https:\u002F\u002Fdiscord.gg\u002FMkH4qj2Jp9) 上的活跃 AGI 社区，一起参与讨论！您的反馈与贡献对我们塑造 big-AGI 的未来至关重要。现在，让我们携手踏上这段激动人心的旅程，共同迎接 `Beam` 🚀","2024-04-03T02:49:26",{"id":209,"version":210,"summary_zh":211,"released_at":212},352195,"v1.14.0","# 🚀 big-AGI 1.14.0 - 模型变形者\n\n### 智能与性能的新形态。\n\n我们发布了 big-AGI 1.14.0 版本，代号“模型变形者”。此次更新扩展了对更多模型的支持，重点提升了性能，并优化了用户体验。本次更新包含超过 230 次提交、修改了超过 8,000 行代码，旨在让 big-AGI 更智能、更快、更可靠。\n\n![release-cover-v1 14 0 b](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fassets\u002F32999\u002F0763d3c0-d165-4978-b45f-37c6e4b79b41)\n\n- (1.14.1) 新增对 **Anthropic Claude-3* 模型的支持\n- 新增对 **Perplexity** 的支持，包括 Sonar Online 模型，以及对 **Groq** 的支持，涵盖最高可达 500 tok\u002Fs 速度的模型。感谢 @Penagwin 的贡献。\n- 新增 **Mistral** Large 模型和 Google **Gemini 1.5** 模型。\n- 深度整合 **LocalAI**，支持模型库和密钥管理。\n- 性能 **优化**：全面改进运行速度、降低功耗并优化内存使用。\n- UI 与易用性：更新了导航、搜索、缩放、快捷键及聊天交互，带来更流畅的体验。\n- 修复了图像警告、独立 LaTeX 块显示问题、人物角色选择器中隐藏角色等若干问题。\n\n### 针对开发者：\n- 可通过环境变量自定义 PlantUML 服务器，感谢 @fredliubojin 的贡献。\n- 现已新增定制化文档和分析支持。\n- 增加了 \u002Finfo\u002Fdebug 接口，便于诊断。\n- 引入基于控制器的聊天架构，简化每轮对话的数据管理，包括临时数据。\n- 完整变更日志：https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fcompare\u002Fv1.13.1...v1.14.0\n\n### 新晋贡献者\n\n特别感谢新加入的贡献者 @Penagwin 和 @jacksongoode。@Penagwin 的工作对本次版本的成功起到了关键作用。\n\n## 总结\n\n1.14.0 版本——“模型变形者”，标志着我们在提供先进、可靠且以用户为中心的人工智能工具方面迈出了重要一步。我们始终坚持去繁从简，聚焦于真正对用户和开发者有价值的内容。\n\n立即探索新功能与改进，升级至“模型变形者”吧！可通过源码、Docker 或直接访问 [big-agi.com](https:\u002F\u002Fbig-agi.com) 获取。\n\n欢迎加入我们的 AGI 社区 Discord 群组（[Discord](https:\u002F\u002Fdiscord.gg\u002FMkH4qj2Jp9)），与我们一起参与开发！您的反馈与贡献是我们不断前进的动力。现在就尽情享受 `Modelmorphic` 吧 🚀","2024-03-06T07:05:36",{"id":214,"version":215,"summary_zh":216,"released_at":217},352196,"v1.13.0","# 🚀 big-AGI 1.13.0 - 多任务 + 个性化\n\n### 更少时间，更多AI。\n\n今天，我们宣布发布 big-AGI 1.13.0，这标志着我们在提升AI交互与生产力的道路上又迈出了重要一步。\n`多任务 + 个性化`聚焦于多任务处理能力和个性化定制，确保您在使用 big-AGI 时更加高效、贴合您的需求。以下是本次更新的新特性：\n\n![img-EvBk3F6tejzYJ81niD6eh5PY](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fassets\u002F32999\u002F22938c52-4edf-4635-9a50-a89f2f6668a6)\n\n- **并排分割窗口**：支持并行对话，实现多任务处理——无需再等待一个助手完成任务 #208  \n- **多聊天模式**：可同时向多个聊天会话发送相同消息 #388 \n- 导出表格为CSV：感谢 @aj47 的贡献，现在您可以将数据表格导出至其他软件，这对于数据分析和报告工作来说无疑是一大突破 #392 \n- **可调节文本大小**：并非一刀切的设计，现在您可以根据需要调整文本大小，尤其是在手机端享受更紧凑的聊天体验 #399 \n- Dev2 人格技术预览：我们正开始测试全新且功能更强大的人格模型\n- UI：聊天界面焕然一新，间距、字体和菜单均得到优化\n- 其他新增内容：集成全新视频播放器、由 @aj47 制作的 [LM Studio 教程](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=MqXzxVokMDk)、[MongoDB 支持](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fblob\u002Fmain\u002Fdocs\u002Fconfig-database.md)（@ranfysvalle02 贡献），以及显著的性能提升\n\n### 针对开发者：\n- 本次发布包含超过150次提交，重点在于内部代码清理与质量提升，同时也带来了几项令人惊艳的新功能。\n- 深入了解详情：[v1.12.0...v1.13.0](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fcompare\u002Fv1.12.0...v1.13.0)\n\n### 新晋贡献者\n我们非常高兴地欢迎 @aj47 和 @ranfysvalle02 加入社区：\n- @aj47 的首次贡献体现在 [#392](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fpull\u002F392)\n- @ranfysvalle02 的首次贡献则是在 [#393](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fpull\u002F393)\n\n---\n\nhttps:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fassets\u002F32999\u002F01732528-730e-41dc-adc7-511385686b13\n\n---\n\n### 立即升级\n别再犹豫，快来体验这些全新功能吧！立即升级至 big-AGI 1.13.0，您可以通过源码、Docker 或直接在托管站点上获取。\n\n加入我们的 AGI 社区 [Discord](https:\u002F\u002Fdiscord.gg\u002FMkH4qj2Jp9)，与我们一起共创未来！您的反馈与贡献是我们不断前进的动力。现在就尽情享受 `多任务 + 个性化` 吧 🚀","2024-02-08T05:50:27",{"id":219,"version":220,"summary_zh":221,"released_at":222},352197,"v1.12.0","# 🚀 big-AGI 1.12.0 - AGI Hotline\r\n\r\nIntroducing big-AGI 1.12.0, a significant step in refining AGI interaction and management. Who can it be now? Your phone is ringing - it's your AGI. **Voice Call**, with **new OpenAI models**, flying on a snappier and **juicier** UI.\r\n\r\n![release-cover-v1 12 0](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fassets\u002F32999\u002F0cad8f75-5751-4471-9bfe-18f1d296811a)\r\n\r\nHere's the essence of what's new:\r\n- **Voice Calls**: real-time voice call your personas out of the blue or in relation to a chat #354 \r\n- **Updated OpenAI 0125** Models. #364 \r\n- Rename and **Auto-Rename** chats.  #222 \r\n- **Link Sharing Revamped**, with more user control #356 \r\n- **Accessibility** to screen readers #358\r\n- Export chats to Markdown #337 \r\n- Paste tables from Excel #286 \r\n- Ollama model updates and context window detection fixes #309 \r\n\r\nFor Developers:\r\n- Over 150 commits and 7,000+ lines changed, emphasizing performance optimizations and under-the-hood improvements for a smoother experience.\r\n- Dive into the details: [v1.11.0...v1.12.0](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fcompare\u002Fv1.11.0...v1.12.0)\r\n\r\nhttps:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fassets\u002F32999\u002F95ceb03c-945d-4fdd-9a9f-3317beb54f3f\r\n\r\n### Upgrade Now\r\nUpgrade to big-AGI 1.12.0 today (code, docker, or the hosted site).\r\n\r\nJoin us on [Discord](https:\u002F\u002Fdiscord.gg\u002FMkH4qj2Jp9) and explore our [public roadmap](https:\u002F\u002Fgithub.com\u002Fusers\u002Fenricoros\u002Fprojects\u002F4\u002Fviews\u002F2). Together, we're shaping its future! 🚀","2024-01-26T12:55:49",{"id":224,"version":225,"summary_zh":226,"released_at":227},352204,"v1.7.1","# 🚀 big-AGI 1.7.1 \r\n\r\n1.7.1 is a small patch release that dramatically improves Ollama chats.\r\n\r\nBy using the [new Chat endpoint](https:\u002F\u002Fgithub.com\u002Fjmorganca\u002Follama\u002Fpull\u002F1392), conversations get optimal translation into prompts, and the result is a great chatting experience across models. You should try out `starling-lm` for development and `neural-chat` for general-purpose chats.\r\n\r\n - **Full Changelog**: https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-agi\u002Fcompare\u002Fv1.7.0...v1.7.1\r\n \r\n![big-AGI_feature-ollama_20231210T230727](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fassets\u002F32999\u002F501f4ae2-11e9-4934-bd88-d7d9fb0fa155)","2023-12-11T06:24:21",{"id":229,"version":230,"summary_zh":231,"released_at":232},352198,"v1.11.0","# 🚀 big-AGI 1.11.0 - Singularity\r\n\r\nLike a cosmic clockwork, today marks the release of big-AGI 1.11.0, and here's a constellation of enhancements that bring the universe of AI closer.\r\n\r\n- **Search conversations**: find by keywords within titles and contents, with frequency ranking - thanks to @joriskalz's stellar work #329 \r\n- **Command palette**: type \"\u002F\" for command selection and auto-completion\r\n- **[Together AI](https:\u002F\u002Ftogether.ai) Integration**: connect with the recent and blazing-fast models provided by the Together AI inference platform #346 \r\n- Persona Creator Enhancements: Deleteable history of creation, custom creation commands, and much more #301 #328 #199 #336 \r\n- OpenAI-Compatible Endpoints: Multiply your AI potential by adding up to five custom OpenAI-compatible endpoints. Flexibility is the new frontier. [#323](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F323)\r\n\r\n1.11.0 for Developers:\r\n- New 'Actiles' framework for developers, to support commands auto-completion with abstract triggers and item sources.\r\n- Full Changelog: [v1.10.0...v1.11.0](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fcompare\u002Fv1.10.0...v1.11.0)\r\n\r\nhttps:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fassets\u002F1590910\u002Fa6b8e172-0726-4b03-a5e5-10cfcb110c68\r\n\r\n### Acknowledgements\r\nThanks especially to @joriskalz for their impactful pull requests in this release.\r\n\r\n### Upgrade Now\r\nUpgrade to big-AGI 1.11.0 today (code, docker, or the hosted site).\r\n\r\nJoin the conversation on [Discord](https:\u002F\u002Fdiscord.gg\u002FMkH4qj2Jp9), explore our [public roadmap](https:\u002F\u002Fgithub.com\u002Fusers\u002Fenricoros\u002Fprojects\u002F4\u002Fviews\u002F2), and thank you for your continued support, and let's explore the possibilities of AGI, together! 🌌🚀","2024-01-16T06:26:52",{"id":234,"version":235,"summary_zh":236,"released_at":237},352199,"v1.10.0","# 🚀 big-AGI 1.10.0 - The Year of AGI\r\n\r\nHappy 2024! We're excited to give you Big-AGI 1.10.0, which brings good improvements and sets the stage a more powerful UX.\r\n\r\nhttps:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fassets\u002F32999\u002Ffbb1be49-5c38-49c8-86fa-3705700f6c39\r\n\r\n- **New UI**: Responsive, animated, and sleek interface for the best UX across desktop and mobile. #201 \r\n- **Conversation Folders**: Keep your chats organized with the new folder system. #321 \r\n- **LM Studio**: model support and improved token management.\r\n- Resizable panes in split-screen chat mode\r\n- Large performance optimizations making big-AGI even faster and more responsive\r\n\r\n1.10.0 for Developers:\r\n - Introducing the new \"OptimaLayout\" UI framework to enable scale and support multiple top-level apps\r\n - Updated the documentation for proxy settings on browserless\u002Fdocker environments. #318\r\n - Full Changelog: [v1.9.0...v1.10.0](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fcompare\u002Fv1.9.0...v1.10.0)\r\n\r\n### New Contributors\r\nWe're excited to welcome @kursad-k to the big-AGI contributors! Thank you, @kursad-k! 🌟\r\n\r\n### Acknowledgements\r\nThanks to @joriskalz for the masterful work on the conversation folders - _try that drag and drop to reorder!_. This community is what makes open source amazing!\r\n\r\n### Support Big-AGI\r\nJoin the conversation on [Discord](https:\u002F\u002Fdiscord.gg\u002FMkH4qj2Jp9), explore our [public roadmap](https:\u002F\u002Fgithub.com\u002Fusers\u002Fenricoros\u002Fprojects\u002F4\u002Fviews\u002F2), and thank you for your continued support, and here's to a fantastic year ahead with big-AGI! 🎊✨","2024-01-06T07:26:24",{"id":239,"version":240,"summary_zh":241,"released_at":242},352200,"v1.9.0","# 🚀 big-AGI 1.9.0 - Creative Horizons\r\n\r\nbig-AGI 1.9.0 paves the way for more creativity and utility. Devs get bigger and better frameworks.\r\n\r\n![DALL·E 3 Example](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fassets\u002F32999\u002Ffd66221e-c301-4db5-a2f2-5fefd99eb570)\r\n\r\n- **DALL·E 3 Support**: includes advanced options, such as resolution, HD, bypass prompt rewrite. [#212](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F212)\r\n- **Persona Creation from Text**: craft AI system prompts from unstructured text (thanks to @joriskalz ). [#287](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fpull\u002F287)\r\n- **Refined Scrolling** Mechanics: responsive scrolling with 'snap to bottom' on all devices. [#304](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F304)\r\n- Openrouter updates for better model management and rate limit handling\r\n- Image drawing UX improvements\r\n- Layout fix for Firefox users #255\r\n\r\n1.9.0 for Developers:\r\n\r\n- New subsystems Optima layout subsystem (Pluggable layouting), Text2Image (swappable image generation providers), ScrollToBottom (in-house optimized scrolling and snap-to-bottom library)\r\n- Improved subsystems: use `react-resizable-panels` for the Panes subsystem, improved LLms abstractions, going towards functions streaming\r\n-Full Changelog: [v1.8.0...v1.9.0](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fcompare\u002Fv1.8.0...v1.9.0)\r\n\r\n## Engage with Us\r\n\r\nJoin the conversation on [Discord](https:\u002F\u002Fdiscord.gg\u002FMkH4qj2Jp9), explore our [public roadmap](https:\u002F\u002Fgithub.com\u002Fusers\u002Fenricoros\u002Fprojects\u002F4\u002Fviews\u002F2), and contribute to the ongoing development of big-AGI. 🌟","2023-12-28T23:02:51",{"id":244,"version":245,"summary_zh":246,"released_at":247},352201,"v1.8.0","# 🚀 **big-AGI 1.8.0 - To The Moon And Back**\r\n\r\nBig-AGI 1.8.0 introduces new integrations and user experience improvements: \r\n\r\nhttps:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fassets\u002F32999\u002F4f397d51-952c-4f41-bcfe-8a935e9988e9\r\n\r\n- **Gemini models** support - try the latest Gemini Pro from Google. [#275](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-agi\u002Fissues\u002F275)\r\n- **Mistral Platform** - added support for Mistral AI models. [#273](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-agi\u002Fissues\u002F273)\r\n- Improved models:\r\n  - **Improved Ollama** chats. [#270](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-agi\u002Fissues\u002F270)\r\n  - Latest versions of Ollama (v0.1.17) and OpenRouter models\r\n- UX improvements:\r\n  - Custom **diagram instructions** by @joriskalz. [#280](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-agi\u002Fpull\u002F280)\r\n  - **Mac Shortcuts Fix**: improve UX for Mac users\r\n  - **Single-Tab Mode** for data integrity. [#268](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-agi\u002Fissues\u002F268)\r\n\r\n1.8.0 for Developers:\r\n\r\n- [**Network troubleshooting**](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fissues\u002F276#issuecomment-1858591483) documentation\r\n- LLMs streaming framework enhancement for future-proofing\r\n- Fixed 1-button Vercel deployment\r\n- Full Changelog: [v1.7.0...v1.8.0](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-agi\u002Fcompare\u002Fv1.7.0...v1.8.0) - 98 files changed\r\n\r\n## New Contributors:\r\n\r\n- A round of applause for @joriskalz for their first contribution to the diagram instructions feature. [PR #280](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fpull\u002F280)\r\n\r\n## Say Hi\r\n\r\nJoin us on [Discord](https:\u002F\u002Fdiscord.gg\u002FMkH4qj2Jp9), check out our [public roadmap](https:\u002F\u002Fgithub.com\u002Fusers\u002Fenricoros\u002Fprojects\u002F4\u002Fviews\u002F2), and help us make big-AGI more useful for everyone. 🌟","2023-12-20T10:39:51",{"id":249,"version":250,"summary_zh":251,"released_at":252},352202,"v1.7.3","# 🚀 big-AGI 1.7.3\r\n\r\n1.7.3 is a small patch release that adds support for the newly announced [Mistral Platform](https:\u002F\u002Fdocs.mistral.ai\u002Fplatform\u002Foverview)\r\n\r\nIn the video below, big-AGI connects to the platform, lists the models, and the uses the 'small' for chat completion (streaming) and prompt generation (non-streaming)\r\n \r\nhttps:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-AGI\u002Fassets\u002F32999\u002F4ddd439c-1e06-4446-bc12-7dac8230828e\r\n","2023-12-13T01:03:39",{"id":254,"version":255,"summary_zh":256,"released_at":257},352203,"v1.7.2","# 🚀 big-AGI 1.7.2\r\n\r\n1.7.2 is a small patch release that updates the OpenRouter models.\r\n\r\nYou have now access to [Mixtral 8x7B](https:\u002F\u002Fmistral.ai\u002Fnews\u002Fmixtral-of-experts\u002F) and for FREE, via OpenRouter 🤑 \r\n - Full Changelog: https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-agi\u002Fcompare\u002Fv1.7.1...v1.7.2\r\n\r\nIn the video below, big-AGI switches from GPT 4 Turbo to Mixtral in a Complex coding task. And the performance is really good compared to what we've seen before.\r\n","2023-12-12T05:30:40",{"id":259,"version":260,"summary_zh":261,"released_at":262},352205,"v1.7.0","# 🚀 **big-AGI 1.7.0 - Attachment Theory**\r\n\r\n1.7.0 rolls in with a game-changing attachment system and more, just in time for the holidays. 🌐✨\r\n\r\n- **Attachments System Overhaul**: the attachments system is brand new. You can now attach files in more than 6 ways, ensuring flexibility and efficiency in bringing content into your conversations:\r\n  - Open files directly\r\n  - Drag and drop from File Explorer or IDE\r\n  - Paste screenshots or copied images\r\n  - Snap photos using the webcam (Labs feature)\r\n  - Use the 'paste' button (Shift + Ctrl + V) for quick clipboard access\r\n  - Paste URLs for automatic embedding\r\n- **Intelligent File Conversion**: The new system converts files to the optimal format for your active Chat LLM, enhancing the chat experience with various content types - for instance:\r\n  - PDFs are automatically converted to text\r\n  - Images undergo OCR for text extraction (for non-vision networks)\r\n  - HTML tables are transformed into Markdown\r\n  - HTML can be as-is or text\r\n  - Text remains as text within Markdown blocks for consistency\r\n- **Desktop Webcam** access for direct image uploads. (Labs) The image is OCR'd (for now), and soon fun modes will be enabled with Vision\r\n- **Overheat LLMs**, raising the maximum model 'temperature' to 2 (can lead to instability, but can also be very creative)\r\n- Oobabooga\u002FText-Generation-WebUI: fix support\r\n- Ollama: update to the latest models\r\n- Quick model settings access via `Ctrl+Shift+O`\r\n- Optimized voice input and app performance\r\n- Cosmetic improvements to the UI\r\n\r\n1.7.0 for Developers:\r\n\r\n- **Password Protection** via HTTP Basic Authentication. [Learn How](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-agi\u002Fblob\u002Fmain\u002Fdocs\u002Fdeploy-authentication.md) :rocket:\r\n- **Browserless Support**: Full browsing support on *Browserless* docker containers [Learn More](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-agi\u002Fblob\u002Fmain\u002Fdocs\u002Fconfig-browse.md) :rocket:\r\n- Full Changelog: https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-agi\u002Fcompare\u002Fv1.6.0...v1.7.0\r\n\r\n## New Contributors:\r\n\r\n- A special shoutout to @konsila \u002F @LennardSchwarz for their first contribution which helped to add HTTP Basic Auth. [PR #177](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-agi\u002Fpull\u002F177)\r\n\r\n## Stay Connected:\r\n\r\nJoin the conversation on [Discord](https:\u002F\u002Fdiscord.gg\u002FMkH4qj2Jp9) or [take a look at the roadmap](https:\u002F\u002Fgithub.com\u002Fusers\u002Fenricoros\u002Fprojects\u002F4\u002Fviews\u002F2) and join us in creating the Pro AGI toolkit ❤️","2023-12-10T12:15:35",{"id":264,"version":265,"summary_zh":266,"released_at":267},352206,"v1.6.0","# 🚀 **big-AGI 1.6.0 - Surf's Up!**\r\n\r\nThe latest update comes with a crowd favorite: browsing, and just a week after our last large 1.5.0 update 🌐✨\r\n\r\n- **Web Browsing**: fetch web pages from \u002Freact, with \u002Fbrowse, or paste a URL - [learn more](https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-agi\u002Fblob\u002Fmain\u002Fdocs\u002Fconfig-browse.md)\r\n- **Branching Discussions**: branch out from any message\r\n- **Keyboard Navigation**: do not leave your keyboard for the mouse (e.g. navigate chats using Ctrl+Shift+Left\u002FRight)\r\n- Performance Boost: large rendering optimizations\r\n- New Features & Tools: use Anthropic Claude 2.1, discover the `\u002Fhelp` command, easily import ChatGPT chats, and streamline your work with the new Flattener\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-agi\u002Fcompare\u002Fv1.5.0...v1.6.0\r\n\r\n1.6.0 for Developers:\r\n\r\n- Improved Code Quality: large refactorings to numerous subsystems\r\n- New frameworks: the Snackbar notification framework, a new streaming hooks\r\n\r\n\r\n## New Contributors\r\n* @g1ibby made their first contribution in https:\u002F\u002Fgithub.com\u002Fenricoros\u002Fbig-agi\u002Fpull\u002F240\r\n\r\n## New Roadmap\r\nWith the new [public open roadmap](https:\u002F\u002Fgithub.com\u002Fusers\u002Fenricoros\u002Fprojects\u002F4\u002Fviews\u002F2), you get a transparent view.\r\nYou can get involved by voting on features, proposing new ideas, and contributing to the development of the app.\r\n\r\nPlease dive into big-AGI 1.6.0! 🏄‍♂️🌊","2023-11-28T09:48:28"]