Skip to content
Get packs

Guide · Open source

How to use NextChat: web app, desktop install and self-hosting

Updated · written with AI from the project's own documentation and checked against our catalogue data

NextChat is an open-source chat interface for AI language models such as GPT, Claude, Gemini and DeepSeek. It runs in a browser as a web app, installs as a desktop app on Windows, macOS and Linux, and can also be self-hosted. This guide explains the main ways to start using it, what the project offers, and what to check before you rely on it.

NextChatChatGPTNextWeb/NextChat · ★ 88.9k · MIT

Quick start

  1. 1Choose your access route. Pick the hosted web app, a desktop install, or self-hosting based on whether you want setup work.
  2. 2Open the web app or desktop app. Use the web app link from the README or download the desktop app from the project's releases page.
  3. 3Clone the repository for self-hosting. Run git clone https://github.com/ChatGPTNextWeb/NextChat to get the source code.
  4. 4Deploy with one click. Use a Vercel or other deploy button and set OPENAI_API_KEY and CODE as prompted.
  5. 5Add your model connections. Enter the API keys for the models you use, or point NextChat at a self-deployed model service.
  6. 6Start chatting and save templates. Chat with your chosen model and create masks for prompts you reuse.

What NextChat is and who it suits

NextChat gives you one chat window for several AI model providers. The project is MIT licensed and written in TypeScript. Its README lists Web, iOS, macOS, Android, Linux and Windows as platforms, and the iOS app is available on the App Store.

  • One chat window for several AI models
  • A self-hosted chat front end for developers who bring their own API keys
  • Browser-based local storage of chat data
  • An enterprise edition described in the README for private deployment

Start with the web app or the desktop app

The simplest route is the hosted web app at app.nextchat.club, which the README links as a Web App Demo. If you prefer a desktop program, the README links a Desktop App page under the project's GitHub releases for Windows, macOS and Linux. The README describes the desktop client as compact, at about 5MB.

The README also says that all data is stored locally in the browser, so chat history stays on the device you use for the web app.

Self-host NextChat by cloning the repository

Developers who want to run their own copy can clone the source code. The README gives this command:

The README lists one-click deploy buttons for Vercel, Zeabur, Gitpod and other services. The Vercel button asks for an OPENAI_API_KEY and a CODE value, where CODE is the page password. Notes in the README say that a one-click deploy creates a new project instead of a fork, so updates may not be detected. It recommends deleting the original repository and using the fork button to redeploy.

  • Use the Vercel one-click button for the quickest deployment
  • Set OPENAI_API_KEY and CODE (your page password) during setup
  • For MCP support, the README says to set ENABLE_MCP=true before building
git clone https://github.com/ChatGPTNextWeb/NextChat

Connect models and your own API keys

The README's Get Started section begins by asking you to obtain an OpenAI API key. The project also states it is fully compatible with self-deployed LLMs and recommends RWKV-Runner or LocalAI for that purpose.

The marketing blurb says no API key is needed to sign up and chat. Read the hosted service's own terms for how that works, and check what you pay for before you rely on it, since the README mentions pay-as-you-go pricing and a sponsor platform.

  • Hosted models: the README names GPT, Claude, Gemini and DeepSeek
  • Self-deployed models: RWKV-Runner or LocalAI are recommended in the README
  • Azure endpoints are supported according to the release notes

Use the main features

The README lists Markdown support with LaTeX, mermaid and code highlighting, plus dark mode, a PWA setup and streaming responses. Prompt templates, called masks, let you create and share chat tools. The project also includes a set of awesome prompts and automatically compresses long chat history to save tokens.

Plugins, Artifacts and Realtime Chat appear in the README's roadmap and release notes. The interface is available in many languages, including English, Deutsch, Français, Español and 日本語.

  • Markdown, LaTeX, mermaid and code highlighting
  • Prompt templates (masks) to create and share chat tools
  • Automatic compression of long chat histories
  • Interface languages including English, 简体中文 and Deutsch

Know the limits

NextChat is built around connecting to hosted or self-deployed model services, so it is not a fully offline assistant. The iOS source code is listed as coming soon in the README. The local knowledge base item is still unchecked on the roadmap.

  • Not an offline assistant on its own
  • iOS source code is marked as coming soon
  • Local knowledge base is listed as not yet done

What people use it for

  • Compare answers from GPT, Claude, Gemini and DeepSeek in one window
  • Run a private chat front end for your own API keys on Vercel
  • Connect NextChat to a self-deployed model through RWKV-Runner or LocalAI
  • Save reusable prompt templates (masks) for recurring writing tasks

Frequently asked questions

Is NextChat free?

NextChat's code is MIT licensed. The README also mentions pay-as-you-go pricing and a sponsor platform, so check the hosted service's terms for any costs before you rely on it.

Do I need an API key to use NextChat?

The marketing text says no API key is needed to sign up and chat. The README's self-hosting steps ask for an OpenAI API key and a CODE page password, so self-hosting requires your own keys.

What platforms does NextChat run on?

The README lists Web, iOS, macOS, Android, Linux and Windows. The desktop app is linked for Windows, macOS and Linux, and the iOS app is on the App Store.

Can NextChat run offline?

Not as a standalone assistant. The README centers on connecting to hosted or self-deployed model services, so you need a model service to chat.

How is NextChat different from Open WebUI or LibreChat?

Comparable projects include LibreChat, Open WebUI, Chatbox and Lobe Chat. This guide does not benchmark them, so compare features such as deployment, model support and data storage against your own needs.

Where is my chat data stored?

The README says the project stores all data locally in the browser. Clearing browser storage or using a different browser may therefore affect your chat history.

New agent skills, repos and AI tools — every week

The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.

Our verdict

NextChat suits people who want one chat window for several AI models, and developers who want a self-hosted front end for their own API keys. Choose something else if you need a fully offline assistant or if you want guaranteed no-cost hosting, since the README points to pay-as-you-go pricing and a sponsor platform.

Install options & downloads