Discover and explore top open-source AI tools and projects—updated daily.
1e0nOpenAI-compatible API proxy for unified LLM access
New!
Top 94.9% on SourcePulse
Summary
droid2api is an OpenAI-compatible API proxy server designed to unify access to diverse Large Language Models (LLMs). It targets developers seeking a single interface for multiple AI providers, offering robust authentication, intelligent inference control, and seamless integration with tools like Claude Code, thereby simplifying complex LLM orchestration.
How It Works
The project functions as a Node.js-based proxy, abstracting LLM provider differences. Core features include a prioritized, multi-factor authentication system (API key, refresh token, client auth) with automatic token renewal. It intelligently manages LLM inference parameters via five "reasoning" levels (auto, off, low, medium, high), dynamically adjusting settings for OpenAI and Anthropic models. The server exposes a standard OpenAI API endpoint, handling automatic request/response format conversion and streaming.
Quick Start & Requirements
npm installnpm startdocker-compose up -dconfig.json for model definitions and ~/.factory/auth.json or environment variables (FACTORY_API_KEY, DROID_REFRESH_KEY) for authentication.Highlighted Details
auto, off, low, medium, high) allow fine-grained management of LLM inference effort and token usage.stream parameter for real-time or batched responses.Maintenance & Community
A QQ discussion group (ID: 824743643) is available for user support and discussion. No other community or maintenance signals are detailed in the README.
Licensing & Compatibility
Licensed under the permissive MIT License, allowing for broad compatibility with commercial and closed-source applications.
Limitations & Caveats
Requires Node.js/npm for local setup. Configuration relies on environment variables or specific JSON files. The primary community channel is a QQ group, which may limit accessibility for some users. Performance claims are not benchmarked.
1 week ago
Inactive
songquanpeng