What you get
Observability
See cost, usage, and performance by agent, feature, and customer.
Rate limits
Request-rate and concurrent-stream limits enforced at the edge.
Any provider, your keys
Works with any OpenAI-compatible app. Bring your own provider keys. They are forwarded untouched and never stored.
Zero infra
No servers, no databases. Toolken runs at the edge. You change a URL and you are done.
Who it’s for
Teams running AI agents (OpenClaw, Hermes, or any framework) and SaaS products that need cost per customer.Start here
Quickstart
Make your first attributed request in under two minutes.
How Toolken works
See the request flow and what is possible.