ChatGPT-Next-Web (NextChat) is a lightweight and fast AI chat assistant that supports multiple models such as Claude, DeepSeek, GPT-4, and Gemini Pro. It provides an iOS app, web app, desktop app, and enterprise edition, and can use AI APIs through pay-as-you-go platforms such as 302.AI, allowing users to register and immediately chat with more than 100 models including GPT-5, Claude 4, Gemini 2.5, and DeepSeek without providing their own API key.
In terms of main capabilities, NextChat provides a compact client of about 5MB on Linux/Windows/macOS, is fully compatible with self-deployed LLMs, and is recommended for use with RWKV-Runner or LocalAI. It insists on privacy first, stores all data locally in the browser, and supports Markdown including LaTeX, mermaid, and code highlighting, with responsive design, dark mode, and PWA, first-screen loading speed of about 100kb, and streaming responses. Version v2 adds the ability to create, share, and debug chat tools through prompt templates (masks), and includes rich prompts provided by awesome-chatgpt-prompts-zh and awesome-chatgpt-prompts. It can automatically compress chat history to support long conversations and save tokens, and supports multiple languages such as English, Simplified Chinese, Traditional Chinese, Japanese, French, Spanish, Italian, Turkish, German, Vietnamese, Russian, Czech, Korean, and Indonesian. The system prompt can fix a user-defined prompt as the system prompt, the user prompt allows users to edit and save custom prompts to the prompt list, and prompt templates can create new chats with predefined context prompts. In addition, it supports sharing as images or sharing to ShareGPT, provides a desktop app based on tauri, and self-hosted models are fully compatible with RWKV-Runner as well as LocalAI server deployments (llama/gpt4all/rwkv/vicuna/koala/gpt4all-j/cerebras/falcon/dolly, etc.). The Artifacts feature makes it easy to preview, copy, and share generated content or web pages through a separate window; plugins support web search, calculators, and other APIs; it also supports Realtime Chat and local knowledge.knowledge base.
In terms of usage, the project supports one-click free deployment on Vercel, completed within 1 minute. During deployment, environment variables need to be set, such as CODE (optional, access password, comma-separated), OPENAI_API_KEY (required), BASE_URL (optional), etc. It also supports API Key and URL configuration for multiple service providers such as Azure, Google Gemini Pro, Anthropic Claude, Baidu, ByteDance, Alibaba Cloud, iFlytek, ChatGLM, DeepSeek, SiliconFlow, 302.AI, etc. Other environment variables include HIDE_USER_API_KEY, DISABLE_GPT4, ENABLE_BALANCE_QUERY, DISABLE_FAST_LINK, CUSTOM_MODELS, DEFAULT_MODEL, VISION_MODELS, WHITE_WEBDAV_ENDPOINTS, DEFAULT_INPUT_TEMPLATE, STABILITY_API_KEY, STABILITY_URL, ENABLE_MCP, etc. Development requires NodeJS >= 18, Docker >= 20. Before local development, you need to create a .env.local file in the project root directory and put the api key in it. It is recommended to use Docker to start the service, and it can also be started behind a proxy. If the proxy requires a password or MCP is enabled, there are corresponding configuration methods. Supports syncing chat records via UpStash.
Applicable scenarios include rapid personal use of multi-model AI chat, self-hosted deployment by developers and enterprises, users who need data privacy and local storage, multilingual team collaboration, and scenarios where internal knowledge bases are integrated for enterprise-level customization and private deployment. The enterprise edition provides brand customization (VI/UI), resource integration (unified configuration management of dozens of AI resources), permission control (member permissions, resource permissions, and knowledge base permissions controlled through the admin panel), knowledge integration (combined with internal knowledge bases), security auditing (automatically intercepting sensitive inquiries and tracing historical conversation records), private deployment (supporting mainstream private cloud solutions), and continuous updates (cutting-edge capabilities such as multimodal AI).
In terms of limitations, the information shows that the project requires users to provide API Keys for some services themselves, and certain features (such as MCP) require setting the ENABLE_MCP=true environment variable before building. The Cloudflare deployment method has been deprecated.If "Updates Available" keeps appearing after one-click deployment, it is because Vercel creates a new project by default instead of forking this repository. You need to delete the original repository, use the fork button to redeploy, and manually enable Workflows and Upstream Sync Action on the Actions page to achieve automatic updates or manually update the code. The source code for the iOS version is "coming soon". For enterprise inquiries, please contact business@nextchat.club. Other information about fees, pricing, platform limitations, or effects not mentioned has not yet been provided.
