RouterFuel: 330+ models, one API.

Connect to 11 model providers and the OpenRouter catalog through one API. Bring your own keys, reuse responses to similar prompts with semantic caching, and switch providers automatically when one fails. RouterFuel is built in Rust and runs on your own server.

Start self-hosting View Documentation

self-hosted via Docker · zero telemetry sent off-cluster · AGPL-3.0 core

build

versionv0.6.1

runtimetokio async

coreaxum · rust

keysBYOK-only

openroutercompatible

licenseAGPL-3.0

Quickstart

postgres + pgvector provisioned automatically

Self-hosted, running in a few commands.

Postgres + pgvector are provisioned automatically by Docker Compose. Nothing to install by hand.

terminal

# 1. Clone and configure
git clone https://github.com/routerfuel/Routerfuel.git
cd Routerfuel
cp .env.example .env   # fill in ROUTERFUEL_ADMIN_KEY at minimum

# 2. Generate a client API key
./scripts/generate-key.sh "MyFirstClient"
# copy the printed hash line into .env's ROUTERFUEL_API_KEYS

# 3. Run it: Postgres + pgvector are provisioned automatically
docker compose up --build

# 4. Make your first request (use the raw key generate-key.sh printed, not the hash)
curl -X POST http://localhost:3000/v1/chat/completions \
  -H "X-API-Key: rf_live_yourgeneratedkey" \
  -H "X-Anthropic-API-Key: sk-ant-yourkey" \
  -H "Content-Type: application/json" \
  -d '{"model":"auto","messages":[{"role":"user","content":"Hello!"}]}'

Gateway Specs

illustrative traffic, not a live feed

routerfuel: example request stream SIMULATED

supported providers

openaidirect

anthropicdirect

geminidirect

deepseek · xai · mistral
qwen · moonshot · zhipu · metadirect

openrouteruniversal fallback

what's shown on the left

This demo shows sample requests to supported providers. It is not connected to a running RouterFuel instance.

Architecture

one binary · no add-on modules

Caching, failover, and spending limits built in.

Every deployment includes these features in the same binary.

4.1

Semantic caching

Checks for similar prompts using a local embedding model. When there is a match, it returns the cached response and skips the provider request.

4.2

Automatic provider failover

Stops sending requests to a provider after repeated errors. Checks whether it has recovered before sending traffic back.

4.3

Loop detection and spending limits

LoopGuard catches clients that keep repeating the same prompt. SpendGuard limits how much each client can spend within a rolling window. Both checks run before the request reaches a provider.

4.4

OpenRouter integration

Use any model in the OpenRouter catalog. RouterFuel refreshes the catalog at startup, so you can access new models without changing your code.

Pricing

free during open beta

Plans based on request volume.

All plans include caching, failover, and spending limits. All features are free during the open beta.

Hobby

For side projects and prototypes.

Freelaunching soon

35,000 requests / month included
Overages billed at $20 / 100k requests
Semantic caching & circuit breaking
Community support

Developer

For apps running in production.

$15/molaunching soon

150,000 requests / month included
Overages billed at $15 / 100k requests
Priority provider failover routing
Email & Slack support

Enterprise

For regulated and air-gapped environments.

Customlaunching soon

Dedicated multi-region deployment
On-premise binary distribution
Custom rate limits
Air-gapped security profiles

Talk to sales

Contact

licensing · deployment · support

Get in touch.

Questions, feedback, or want to talk deployment? Email us directly, or book a call.

contact@routerfuel.com Book a 15-min call

licensing

Need a commercial license?

RouterFuel's core is licensed under AGPL-3.0. That works great for self-hosting and open-source use, but if you're embedding RouterFuel in a closed-source product or run into AGPL's network-use clause at your company, we offer a separate commercial/enterprise license with no copyleft obligations.

Talk to us about licensing