xiry.

Development · GitHub repo

FreeLLMAPI

Route multiple free-tier LLM providers through a single OpenAI-compatible endpoint

From the Xiry library · Updated 3 October 2026

Visit github.com ↗

FreeLLMAPI is an open-source local proxy and router that aggregates free quotas across dozens of AI providers into a single unified endpoint. It automatically handles failover on rate limits, tracks per-key usage, and stores provider credentials with AES-256 encryption. It allows developers to plug free model access directly into coding agents, CLIs, and OpenAI-compatible apps.

What it's for

  • Run coding agents like Claude Code, Codex, Aider, and Cline using free LLM tiers
  • Prevent 429 rate-limit interruptions through automatic provider failover
  • Track token and request limits across dozens of free model APIs automatically
  • Connect self-hosted local models and remote gateways into one unified endpoint

Features

  • OpenAI-compatible endpoints for chat, completions, audio, images, and embeddings
  • Anthropic Messages, native Gemini, and Ollama emulation support
  • Smart routing with automatic fallback on 429 or 5xx provider errors
  • AES-256-GCM encrypted local credential storage in SQLite
  • Automated setup generators for coding tools like Claude Code, Codex, and Aider
  • Web dashboard and desktop menu-bar app with usage analytics and model management
  • MCP server support for direct connection to tools like ChatGPT

Pricing

Freemium. Free and open source with a monthly model catalog snapshot; optional live daily catalog sync is $19/year

Details

Type
GitHub repo
Category
Development
Topic
AI Gateways
Open source
Yes
Platforms
macOS, Windows, Linux, Command line, API, Self-hosted
For
Developers, AI tinkerers, and coding-agent users looking to maximize free AI quotas
Alternative to
LiteLLM, OpenRouter, Portkey
Tags
llm, api router, openai, proxy, free tier, self hosted

Why it was saved

  • @tysn.dev, 6 June 2026: Gives developers and AI agent runners access to nearly 2 billion free tokens per month with automatic failover when individual provider rate limits are hit.

Compare

Related

Updated