Independent tools for your model stackOpen source / MIT licensed

Every model.
One address.

Alnair is an open-source LLM router and AI API gateway. Bring your providers together, set fallback chains, and follow every token from a dashboard that runs on your machine.

Install Alnair Router Read the quick start

Single Rust binary. Embedded dashboard. SQLite storage.

The router deskOn your own terms.

Alnair Overview dashboard showing usage totals and live traffic with sample data — Overview / See the whole desk.

Actual dashboard. Screenshots contain sample data.

Bring your own providersOpenAI + Anthropic wire formatsSelf-hosted by you

01 / The routing desk

Your client keeps its familiar API.

Give your models
a forwarding address.

Point your OpenAI-compatible client at Alnair. A model prefix picks the upstream. A named combo gives you an ordered chain for multi-provider fallback.

Use /v1/chat/completions or /v1/responses for OpenAI clients, and /v1/messages for Anthropic clients. Keep your own provider accounts.

Read the routing rules

A request's routeIllustrated example

Your clientPOST /v1/chat/completions

Alnair Routermodel: "free-forever"

First upstreamNext upstream

If a tier fails before emitting content, the next tier gets the request. You configure the order.

Keep keys in check.: Give client keys rate limits, budgets, model allowlists, and plans.
Know what ran.: Inspect requests, token counts, recorded spend, and provider activity.
Trim bulky context.: Use the token-saving pipeline to compress tool output and inspect the savings.

02 / A little proof of work

Your usage,
as a sticker.

A snapshot worth keeping. Turn your API key's usage into a retro card with total tokens, requests, recorded spend, and your top models.

Download a PNG in landscape or square format. Or open your device's share sheet, where supported.

See more of the dashboard

The exported card leaves out your API key and router address.

My Usage page with a retro usage card, top model breakdown, landscape and square formats, and Download PNG and Share image buttons; sample data — Real export controls in the dashboard. Usage shown here is sample data.

03 / Put it on your desk

Your machine.
Your model stack.

Install the router, open the local dashboard, and connect your first provider.

Alnair is MIT licensed. You bring your provider accounts and pay their usage costs.

Download a release binary

Shell

npm install -g @xflawlessdev/alnair-router

Run alnair-router start to launch the router. To also enable auto-start with your session, run alnair-router install.

Next stop: your dashboard

Run alnair-router start, then open:

http://127.0.0.1:7878/login

Use the setup code from the router log to choose your dashboard password. Then add an upstream connection.

Full install guide, including Docker

Every model.One address.

Give your modelsa forwarding address.

Your usage,as a sticker.

Your machine.Your model stack.

Every model.
One address.

Give your models
a forwarding address.

Your usage,
as a sticker.

Your machine.
Your model stack.