Repolode
Submit
tashfeenahmed

tashfeenahmed/freellmapi

A self-hosted OpenAI-compatible gateway that pools free LLM provider tiers behind one /v1 endpoint, with routing and failover.

Open-source alternative to:OpenRouterLiteLLM

View on GitHubWebsite
Stars
38,952
Forks
5,249
Open issues
40
Last commit
7 hours ago

What is it?

FreeLLMAPI is a local gateway that aggregates the free tiers of many LLM providers and exposes them through a single OpenAI-compatible /v1 endpoint. You add your provider keys in the dashboard, and any OpenAI SDK or compatible client can then send requests to one base URL. A router picks an available model for each request and falls over to the next provider when one returns a rate-limit or server error.

Provider keys are stored encrypted in SQLite and decrypted in memory per request, and the app tracks requests, tokens and per-key rate limits so routing stays under each free-tier cap. It also covers chat, embeddings, image, video and speech endpoints, plus Anthropic Messages, Gemini and Ollama-style surfaces for coding agents and CLIs. The router runs on Node 20+ and ships as a Docker image, a desktop app, and a Premium option that adds a live model catalog feed.

Key features

  • 01Single OpenAI-compatible /v1 endpoint across many free provider keys
  • 02Automatic fallover to the next model on 429 or 5xx responses
  • 03Per-key RPM, RPD, TPM and TPD tracking to stay under free-tier limits
  • 04Provider keys encrypted at rest with AES-256-GCM
  • 05Anthropic, Gemini and Ollama-compatible endpoints for coding agents
  • 06Admin dashboard with key management, playground and usage analytics

Use cases

  • Prototyping an app against many LLMs without paying for API usage
  • Pointing Claude Code, Codex CLI or Aider at a pool of free models
  • Running a personal local gateway that mixes cloud free tiers with self-hosted models
  • Testing how different providers respond to the same OpenAI-style request

Quick start

  1. 1

    Download the desktop app from the releases page (macOS or Windows)

  2. 2

    Self-host with Docker by following the Docker Compose instructions in the install guide

Things to know

  • Free installs receive new models 30 days after they reach the live premium feed, while premium is a paid plan at $19/yr.
  • You must obtain and add your own API keys from each free provider you want to use.

Topics

#llm#openai-compatible#api-gateway#llm-routing#failover#free-tier#self-hosted#typescript

FAQ

What is freellmapi?
A self-hosted OpenAI-compatible gateway that pools free LLM provider tiers behind one /v1 endpoint, with routing and failover. FreeLLMAPI is a local gateway that aggregates the free tiers of many LLM providers and exposes them through a single OpenAI-compatible /v1 endpoint. You add your provider keys in the dashboard, and any OpenAI SDK or compatible client can then send requests to one base URL. A router picks an available model for each request and falls over to the next provider when one returns a rate-limit or server error.
Is freellmapi free to use? What license does it have?
freellmapi is open source under the MIT license. Check the license terms before using it in commercial products.
Is freellmapi still maintained?
Actively maintained. The last commit to freellmapi was on Oct 21, 2026.
What can I use freellmapi for?
Prototyping an app against many LLMs without paying for API usage Pointing Claude Code, Codex CLI or Aider at a pool of free models Running a personal local gateway that mixes cloud free tiers with self-hosted models Testing how different providers respond to the same OpenAI-style request
How do I install freellmapi?
Download the desktop app from the releases page (macOS or Windows) Self-host with Docker by following the Docker Compose instructions in the install guide
Where can I get freellmapi?
The source code and installation instructions are on GitHub: https://github.com/tashfeenahmed/freellmapi

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects