📊 See open-source project rankings

📱 Get the AI Rank app

FreeLLMAPI (GitHub: tashfeenahmed/freellmapi) is a self-hosted router that puts many free LLM provider tiers—and any custom OpenAI-compatible chat, embedding, image, or audio endpoint—behind one /v1 API. The README we opened on November 1, 2026 markets roughly 7.4 billion listed free-tier tokens per month across 34 free providers and 635 free model endpoints, with smart routing, automatic failover, and AES-256-GCM encrypted provider keys. It is MIT-licensed and aimed at personal experimentation, not production SLAs.

What FreeLLMAPI does

Instead of juggling dozens of SDKs and rate limits, you run a local (or Docker) server, add provider keys in the dashboard, and point OpenAI-compatible clients at http://localhost:3001/v1 with a unified freellmapi-… bearer token. The README says the router:

  • Exposes OpenAI-style surfaces (/v1/chat/completions, /v1/responses, embeddings, images, video, speech, and more) plus Anthropic /v1/messages and optional Gemini / Ollama-compatible paths.
  • Scores models for speed/capability/reliability, retries on 429/5xx with cooldowns and key rotation.
  • Syncs a signed model catalog from freellmapi.co twice a day; free installs use a monthly snapshot (models join it 30 days after the live feed), while a paid “Premium” live feed ($19/year or $49 lifetime on that site) updates same-day.
  • Keeps provider keys encrypted at rest in SQLite and decrypts them in memory per request.

Named free providers in the README include Google, Groq, Cerebras, Mistral, OpenRouter, Cloudflare, Cohere, Z.ai (Zhipu), NVIDIA, HuggingFace, ModelScope, and others, plus a custom OpenAI-compatible provider slot.

Docker / one-liner self-host (from the README)

Requirement called out by upstream: Docker for the one-liner path.

curl -fsSL https://freellmapi.co/install.sh | bash

The README says the script sets up ~/freellmapi, generates an encryption key, pulls the image, and starts the container; re-running is safe and preserves .env. Prefer reading the install script before piping to bash. After start, open http://localhost:3001, add keys on Keys, adjust the Fallback Chain, and copy the unified API key from the Keys header.

Windows users are pointed at the desktop .exe installer on GitHub Releases; macOS gets .dmg builds (arm64 / x64). Server-style Docker Compose, local Node 20+ development, LAN access, and backups are documented in docs/en/install/01-install.md. A GHCR Docker image badge is linked from the README (ghcr.io freellmapi).

Minimal OpenAI Python client shape from the README:

from openai import OpenAI
client = OpenAI(base_url="http://localhost:3001/v1", api_key="freellmapi-your-unified-key")
resp = client.chat.completions.create(
    model="auto",
    messages=[{"role": "user", "content": "Summarise the fall of Rome in one sentence."}],
)

Practical caveats (README Limitations + Disclaimer)

Treat these as hard constraints from upstream, not our guesses:

  • Personal experimentation / learning only—not a production inference substrate. Swap to paid APIs before shipping real products.
  • Free tiers mean no frontier models, variable latency, and no SLA. Effective quality can dip late in the day as daily caps are hit, then reset at UTC midnight (README Limitations).
  • Your relationship with each upstream provider still follows their Terms of Service; the project includes a provider-by-provider ToS review note dated May 2026 in its architecture docs.
  • Server installs use an email + password dashboard account; desktop apps auto-sign in locally. Forgot-password codes print to server logs (no email reset).
  • Catalog numbers (providers, endpoints, token budgets) change; re-check freellmapi.co/models and the live README before you rely on a count.

FAQ

What is FreeLLMAPI on GitHub?

It is the open-source repo tashfeenahmed/freellmapi: a local OpenAI-compatible gateway that aggregates free LLM provider tiers with routing and encrypted key storage (MIT).

How do I self-host FreeLLMAPI with Docker?

Use the official one-liner (curl …/install.sh | bash) when you accept piping an install script, or follow Docker Compose / install docs linked from the README. Docker is required for that one-liner path.

Is FreeLLMAPI free?

The router is MIT-licensed and free to run. Upstream sells an optional Premium live catalog feed; free installs still update from a monthly snapshot.

Can I use Claude Code or Codex with it?

The README lists setup helpers such as npx freellmapi setup-claude --url http://localhost:3001 --api-key <unified-key> and documents many OpenAI / Anthropic / Gemini-compatible agents. Follow upstream’s client guide for your tool.

Sources