Getting started
Quickstart
Go from zero to your first routed request in about five minutes.
TokenRouter is one OpenAI-compatible endpoint in front of 13 AI providers, with budgets, rate limits, and analytics enforced at the gateway. This quickstart takes you from a fresh account to your first routed request.
1. Create an account
Sign up and pick a plan — every plan starts with a 14-day free trial. You’ll land in the console with a default organization and team.
2. Add a provider key
TokenRouter is BYOK: you bring your own provider keys and pay providers directly, with zero markup. In the console, go to Provider Keys and add a key for at least one provider (for example your OpenAI or Anthropic key). Keys are AES-encrypted at rest and never logged.
3. Create a TokenRouter API key
Go to API Keys and create a key. Gateway keys look like tr_... and are shown once — store yours somewhere safe. Best practice: one key per app, so usage and budgets attribute cleanly.
4. Make your first request
curl https://api.tokenrouter.io/v1/chat/completions \
-H "Authorization: Bearer tr_your_key_here" \
-H "Content-Type: application/json" \
-d '{
"model": "openai/gpt-5-mini",
"messages": [{"role": "user", "content": "Hello, TokenRouter!"}]
}'That’s the whole integration: point any OpenAI-compatible client at https://api.tokenrouter.io/v1 with your tr_ key. Anthropic clients work too — see the Messages API.
Next steps
- Model IDs and auto-routing — including
autoandauto:cost. - Teams & budgets — hard caps, alerts, and allowlists.
- Use TokenRouter with Claude Code.