FreeLLMAPI

FreeLLMAPI is an open-source router that stacks the free tiers of 34 LLM providers behind a single OpenAI-compatible endpoint — roughly 7.4 billion free tokens per month across 635 model endpoints — with more than 38,000 stars on GitHub.

Overview

Every major AI lab offers a free tier, but each one has its own SDK, rate limits and failure modes. FreeLLMAPI collapses all of them into one self-hosted server: you add your provider keys once, and any OpenAI-compatible client — or Claude Code, Codex, Cursor and dozens of other coding agents — talks to a single /v1 endpoint with one unified key. A smart router scores models live on speed, capability and reliability, fails over to the next provider when one rate-limits, and tracks per-key usage so you stay under every free-tier cap.

Keys are stored AES-256-GCM encrypted, and the router keeps itself current: a signed model catalog syncs automatically, so new free models and quota changes arrive without manual updates. Beyond chat, the same endpoint serves embeddings, image and video generation, speech and transcription, and it also speaks the Anthropic Messages, Gemini and Ollama API formats.

Key Features

• One endpoint, 34 providers – Google, Groq, Cerebras, Mistral, OpenRouter, Cloudflare, Cohere, NVIDIA, Hugging Face and more, plus any custom OpenAI-compatible endpoint.
• Smart routing and failover – six routing strategies with live scoring, automatic retry on rate limits and key rotation.
• Per-key rate tracking – RPM, TPM and daily counters that learn each provider’s ceilings so routing never exceeds a cap.
• Multi-model fusion – fan a prompt out to several free models in parallel and let a judge model synthesize one answer.
• Full API surface – chat, responses, images, video, speech, transcription and embeddings, streaming and non-streaming.
• Encrypted key vault – provider keys encrypted at rest; apps only ever see one unified bearer key.
• Dashboard and desktop app – web admin UI with analytics and playground, plus native macOS and Windows apps and Docker deployment.

💳 Pricing

FreeLLMAPI is free and open source under the MIT license for self-hosting; you supply your own free-tier provider keys. An optional premium tier ($19/year) upgrades the model catalog from monthly snapshots to same-day live updates. The project is intended for personal experimentation.

Who Is It For

Developers, students and hobbyists who want maximum free inference for coding agents and side projects, and who are happy to self-host a small router to get it.