This project provides a free, unified API gateway for various large language models, including GPT, DeepSeek, Claude, Gemini, and Grok. It targets developers and users seeking convenient, high-speed access to multiple LLMs without requiring VPNs or proxies, especially within China. The primary benefit is simplified integration and cost savings through free access tiers.

How It Works

The service acts as a proxy, forwarding requests to official LLM APIs. It utilizes dynamic acceleration for domestic users, aiming for lower latency than direct proxying. The core advantage is its unified OpenAI-standard API, allowing users to switch between different models (e.g., GPT-4o, DeepSeek-R1, Claude-3-5-Sonnet) simply by changing the model name in their requests, streamlining integration with existing applications.

Quick Start & Requirements

To use the service, users must obtain a free API key by binding their GitHub account. The API endpoint can be set to https://api.chatanywhere.tech/v1 (for domestic users) or https://api.chatanywhere.org/v1 (for international users). Configuration involves updating the base_url in OpenAI client libraries or setting the OPENAI_API_BASE environment variable. Detailed integration examples are provided for Python's OpenAI library, gpt_academic, Gomoon, Zotero plugins, BotGem, ChatBox, ChatGPT Sidebar browser extension, and Jetbrains IDE plugins.

Highlighted Details

• Offers free access to models like GPT-4o (5 times/day), DeepSeek-R1/V3 (30 times/day), and GPT-3.5-Turbo (200 times/day).
• Supports streaming responses and is compatible with OpenAI's API standards.
• Provides a paid tier with lower prices than official OpenAI rates for enhanced GPT-4 access and unlimited usage.
• Includes support for image generation (DALL-E) and speech-to-text (Whisper) models.

Maintenance & Community

The project maintains a QQ group (599246660) for community support and announcements. Service availability is monitored and displayed via status badges.

Licensing & Compatibility

The project itself does not specify a license for its code. The free API keys are explicitly for personal, non-commercial, educational, and non-profit research use. Commercial use or large-scale model training with free keys is strictly prohibited.

Limitations & Caveats

Free API keys are subject to rate limits (200 requests/day/IP & Key) and may be temporarily blocked if abused. The project states that its system is for internal evaluation and testing; commercial or public-facing use is at the user's own risk. Some third-party models (e.g., certain Claude, Gemini, Grok versions) may experience occasional slow responses or errors due to reliance on third-party providers.