Metered upstreams
Models, payments, and maps bill every test call. The twin counts what you would have spent.
Stateful API simulation
The twin is a stateful mock of OpenAI, payments, and APIs that are still being built. It fails on purpose, and answers curl from the first minute.
These buttons call the public twin at /sim/demo. Edit the order body, then create it.
Models, payments, and maps bill every test call. The twin counts what you would have spent.
An order created with POST is there on the next GET. Static JSON cannot do that.
Switch the whole dependency to Degraded or Outage and get latency, errors, and 503s on demand.
List, create, read, update, and delete are separate routes that share one collection.
chat/completions returns JSON, SSE, and tool_calls.
Normal, Degraded, and Outage. Latency, error rate, and RPM are configurable.
API keys, curl, and a control API that flips the scenario from a pipeline.
Paste OpenAPI JSON or one real response and get routes.
Emails, card numbers, and SSNs stay out of the request log.
$0
$8/month
$12/month
Talk to us
Point a twin at an upstream. Calls are forwarded and stored, or POST a capture to /__tap and save it as a route.
Unary JSON RPCs, a socket on port 3001, and topics with offsets. The same twin URL serves all three.
Team connects an OIDC issuer to an email domain. Enterprise connects SAML. Those accounts join the workspace as viewers.
You pay when a period starts. Developer is $8 a month or $50 for the year. Team is $12 a month or $120 for the year. Paying clears this month's call count. A yearly plan has no monthly cap. The webhook keeps the workspace on that plan while the subscription is active.