Skip to content

README.md

Latest commit

 

History

History
219 lines (157 loc) · 7.45 KB

README.md

File metadata and controls

219 lines (157 loc) · 7.45 KB
Raw

Chat2API Web

Chat2API Logo

Release License
Node.js React TypeScript Platform

中文

Multi-platform AI Service Unified Management Tool — Web Edition

Chat2API Web enables zero-cost access to leading AI models by leveraging official web UIs. Deploy on any server (VPS, NAS, local machine) and manage everything from a browser. Supports DeepSeek, GLM, Kimi, MiniMax, Perplexity, Qwen, Z.ai and more — making any OpenAI-compatible client work out of the box.

Product Preview

✨ Features

  • Pure Web Architecture — No Electron, no desktop app. Single Node.js process serves UI + API on one port
  • OpenAI Compatible API — Standard /v1/chat/completions endpoint, works with any OpenAI client
  • Multi-Provider Support — DeepSeek, GLM, Kimi, MiniMax, Perplexity, Qwen, Qwen AI, Z.ai, Mimo
  • Context Management — Sliding window, token limit, and summary compression strategies
  • Function Calling — Universal tool calling via prompt engineering, compatible with Cherry Studio, Kilo Code, Cline
  • Model Mapping — Flexible model name mapping with wildcard and preferred provider/account
  • Load Balancing — Round-robin, fill-first, and failover strategies across multiple accounts
  • Session Management — Auto-cleanup of provider chat sessions, batch deletion support
  • API Key Management — Generate and manage keys for proxy authentication
  • Dashboard — Real-time request traffic, latency, and success rate monitoring
  • Request Logs — Detailed logging for debugging and analysis
  • Multilingual — English and Simplified Chinese
  • Dark/Light Theme — Modern responsive UI

🤖 Supported Providers

Provider Auth Type Models Status
DeepSeek User Token deepseek-v4-pro, deepseek-v4-flash Available
GLM Refresh Token GLM-5.2 Available
Kimi JWT Token Kimi-K2.6 Available
MiniMax JWT Token MiniMax-M2.7 Available
Perplexity Session Token Auto (Free) Available
Qwen (CN) SSO Ticket Qwen3.6, Qwen3.7-Max, Qwen3.5-Flash, Qwen3-Max, Qwen3-Coder Available
Qwen AI (Global) JWT Token Qwen3.7-Max, Qwen3.6-Plus, Qwen3.6-35B-A3B, Qwen3.6-27B, Qwen3-Coder Available
Z.ai JWT Token GLM-5.2, GLM-5-Turbo, GLM-5V-Turbo, GLM-5, GLM-4.7 Temporarily unavailable due to frontend captcha risk control
Mimo Service Token MiMo models Available

📥 Installation

Option 1: Docker (Recommended)

The simplest way to deploy. Requires only Docker and Docker Compose.

# Clone the repo
git clone https://github.com/zhaiiker/Chat2API-web.git
cd Chat2API-web

# Build and start in background
docker compose up -d --build

# View logs
docker compose logs -f chat2api

The service will be available at http://your-server:8080.

Custom port:

# Edit docker-compose.yml ports section, or use:
docker compose up -d --build
# Then map host port as needed

To change the exposed port, edit docker-compose.yml:

ports:
  - "9000:8080"  # Map host port 9000 to container port 8080

Environment variables (optional):

Create a .env file in the project root:

CHAT2API_MANAGEMENT_SECRET=your-secret-here
CHAT2API_ENCRYPTION_KEY=your-encryption-key-here

Or generate a secure secret:

openssl rand -hex 32

Data persistence: The docker-compose.yml uses a named volume (chat2api-data) mapped to /data inside the container. All configuration, accounts, and logs are preserved across container restarts and rebuilds.

Common Docker commands:

# Stop the service
docker compose down

# Rebuild after pulling updates
git pull
docker compose up -d --build

# View real-time logs
docker compose logs -f chat2api

# Reset data (caution: deletes all settings)
docker compose down -v

Option 2: Direct Deployment

Requirements: Node.js 20+ (LTS recommended: 22.x), npm, Git

# Clone
git clone https://github.com/zhaiiker/Chat2API-web.git
cd Chat2API-web

# Install dependencies
npm ci

# Build frontend + backend
npm run build

# Start (default port 8080)
PORT=8080 node dist/backend/index.js

Open http://your-server:8080 in a browser. On first visit you'll set an administrator password.

Run with PM2 (recommended for production):

npm install -g pm2

# Start
PORT=8080 pm2 start dist/backend/index.js --name chat2api

# Auto-start on boot
pm2 startup
pm2 save

# View logs
pm2 logs chat2api

# Restart after update
git pull && npm ci && npm run build
pm2 restart chat2api

Option 3: Development Mode

Requirements: Node.js 20+, npm

npm ci
npm run dev

This starts both backend (port 8080) and frontend dev server (port 5173) with hot reload.

⚙️ Configuration

All configuration is done via environment variables or the .env file in the project root:

Variable Default Description
PORT 8080 Server port
HOST 0.0.0.0 Bind address
CHAT2API_DATA_DIR ~/.chat2api Data directory (Docker: /data)
CHAT2API_MANAGEMENT_SECRET (auto-generated) Management API secret (pin for persistence)
CHAT2API_ENCRYPTION_KEY (auto-generated) Credential encryption key (pin for persistence)

Tip: In production, always pin CHAT2API_MANAGEMENT_SECRET and CHAT2API_ENCRYPTION_KEY so they survive container restarts. Auto-generated keys change on each cold start, which would invalidate existing sessions and encrypted credentials.

🔧 Usage

  1. Open the web UI → Add a provider (e.g. DeepSeek)
  2. Paste your token from the provider's website
  3. Create an API Key in the API Key Management page
  4. Point your AI client to http://your-server:8080/v1 with the API key

Works with Cherry Studio, Chatbox, OpenCat, Cline, Roo-Code, or any OpenAI-compatible tool.

🔄 What's New in v1.4.0

  • Model Updates — All providers synced to latest upstream models (DeepSeek V4, GLM-5.2, Qwen3.6/3.7, MiniMax-M2.7, etc.)
  • Session Management — Qwen and Kimi now support batch session listing and deletion
  • DeepSeek Search Enhancement — Improved search result merging, citation handling, and semantic model detection
  • Thinking Mode Improvements — Better thinking/search/fold mode detection across all providers
  • Updated Browser Fingerprints — Chrome 148 UA and headers for DeepSeek, Z.ai

友情链接

  • LINUX DO —— 新的理想型社区，技术爱好者的聚集地。

📄 License

GPL-3.0