Unified AI Gateway

One API key for every model

Connect any AI client, agent, SaaS backend, or workflow to OctaRouter and use the same API key for every model. Choose the protocol that fits: OpenAI-compatible, Anthropic Messages, or Gemini generateContent.

Drop-in OpenAI-compatible call
$ curl https://api.octarouter.com/v1/chat/completions \
  -H "Authorization: Bearer sk-or-..." \
  -H "Content-Type: application/json" \
  -d '{"model":"gpt-4o","messages":[...]}'

Pick a model on the Models page, then follow the official API docs for request parameters.

Why OctaRouter

A lighter way to run model traffic across apps, agents, and backends.

Fewer integrations

One account, one base URL, and one API key instead of signing up, topping up, and integrating with each model provider separately.

Pay as you go

Pay only for the requests you make. Use what you top up — no top-up fee.

Clear visibility

Request logs show model, tokens, cost, and status so you can trace failures fast.

Built for

Teams shipping AI features that need reliable routing and predictable spend.

AI-powered apps

Ship chat, summarization, or code generation through OctaRouter and switch models without opening new provider accounts.

Agents & workflows

Run tool-calling agents with model fallbacks. Use the same OctaRouter API key across protocols in production.

SaaS backends

Meter usage per customer with gateway logs and a shared prepaid pool — configure OctaRouter once in your backend.

Scripts & automation

Use the same API key from curl, CI, or cron — switch models or protocols without managing another account.

How it works

Set up once, then route requests across models with one key.

1

Create account

Sign up, verify your email, and add prepaid credits to your balance.

2

Create an API key

Create a gateway key in the dashboard. You can view and copy the full secret again anytime on the API keys page.

3

Send your first request

Choose a model and protocol from the Models page, then call OctaRouter with the matching base URL and endpoint path.

Three protocols, one API key

The OpenAI-compatible protocol is the default for every model. Claude and Gemini also support their native protocols. See the docs for protocol-specific base URLs and official API references.

Ready to route your first request?

Create an account, add credits, and use the same API key across every available model and supported protocol.

Start routing requests