Native APIs
OpenAI and Anthropic wire formats, including streaming. Keep your existing SDK.
OPEN SOURCE · MIT LICENSED
An open-source
AI model router.
Route OpenAI and Anthropic requests through one self-hosted Rust binary. Manage provider keys, namespace budgets, and usage in one place.
Completely open source. Use it, modify it, and run it on your own infrastructure.
Currently in beta.See the compatibility contract.
Scoped controls
Namespace credentials, period budgets, usage metering, and OpenTelemetry.
Your deployment
A single binary with SQLite or Postgres. Run locally, in a container, or on your servers.
01 / INSTALL
Install with one command.
The installer detects your platform and downloads the prebuilt binary. Then configure your provider key and start Axond.
Configure and run AxondRead the shell script ·PowerShell script
Other installation optionscurl -fsSL https://axond.dev/install.sh | shirm https://axond.dev/install.ps1 | iexDownloads the latest release binary and verifies its SHA-256 checksum. No Rust toolchain or Docker required.
02 / SEND A REQUEST
Use your existing SDK.
Set the base URL to your namespace and use a provider-prefixed model ID, such as openai/gpt-4o.
These examples use the local quickstart key. Configure a real provider key and budget first.
Using Anthropic? See its client guideRequests keep their native wire format. Axond does not translate between OpenAI and Anthropic APIs.
from openai import OpenAIclient = OpenAI( base_url="http://localhost:8080/ns/platform/v1", api_key="quickstart-platform-key",)response = client.chat.completions.create( model="openai/gpt-4o", messages=[{"role": "user", "content": "Hello, Axond."}],)print(response.choices[0].message.content)import OpenAI from "openai";const client = new OpenAI({ baseURL: "http://localhost:8080/ns/platform/v1", apiKey: "quickstart-platform-key",});const response = await client.chat.completions.create({ model: "openai/gpt-4o", messages: [{ role: "user", content: "Hello, Axond." }],});console.log(response.choices[0].message.content);curl http://localhost:8080/ns/platform/v1/chat/completions \ -H "Authorization: Bearer quickstart-platform-key" \ -H "Content-Type: application/json" \ -d '{ "model": "openai/gpt-4o", "messages": [ {"role": "user", "content": "Hello, Axond."} ] }'