Summary

This project offers a proxy API for Qwen Large Models, simplifying integration and management for developers. It targets users needing a flexible backend for Qwen LLMs, providing multi-key authentication, advanced model features, and diverse deployment options to streamline access and control.

How It Works

Qwen-Proxy exposes Qwen models via a RESTful API, largely mirroring the OpenAI specification. It handles request routing and API key authentication, supporting single-process or multi-process deployments (Node.js cluster/PM2) for performance. Data persistence (file/Redis) and caching are available. Novel features like intelligent search (-search) or reasoning output (-thinking) are enabled by appending suffixes to model names.

Quick Start & Requirements

• Installation: Deploy via Docker (docker run, docker-compose), local Node.js (npm install, npm start), or PM2.
• Prerequisites: Node.js 18+ for source deployment; Docker recommended; Redis optional for persistence.
• Resources: Standard dependency installation. Resource usage scales with Qwen models.
• Links: Docker Hub: https://hub.docker.com/r/rfym21/qwen2api; Docker Compose: https://raw.githubusercontent.com/Rfym21/Qwen2API/refs/heads/main/docker-compose.yml; Community: https://t.me/nodejs_project.

Highlighted Details

• Multi-API Key Management: Supports admin/user roles via multiple API keys, enabling granular access control and dynamic management.
• OpenAI API Compatibility: Mimics OpenAI API endpoints (/v1/models, /v1/chat/completions) for easy integration.
• Advanced Model Features: Enables search (-search), thinking (-thinking), and combined modes via model name suffixes.
• Image Generation & Multimodality: Supports text-to-image generation and multimodal input (text with image URLs).
• CLI Endpoint: Dedicated CLI endpoint (/cli/v1/chat/completions) for qwen3-coder-plus/flash models, supporting 256K context and function calling.

Maintenance & Community

A Telegram community group (https://t.me/nodejs_project) is provided. The README lacks details on core maintainers, sponsorships, or a roadmap, but versioning and Docker images suggest active maintenance.

Licensing & Compatibility

The README omits license information, posing a significant adoption risk, especially for commercial use. The API design targets compatibility with OpenAI-like interfaces.

Limitations & Caveats

The primary limitation is the unspecified software license. Source deployment requires Node.js 18+. The CLI endpoint is restricted to specific models (qwen3-coder-plus, qwen3-coder-flash) and lacks the full feature set of the main API.