FASHN

The FASHN API is an asynchronous prediction service. Clients POST to /v1/run with a model_name and model-specific inputs, then poll /v1/status/{id} until the prediction completes. Authentication uses a Bearer token. Rate limits: 50 req/60s on /run, 50 req/10s on /status, 6 concurrent predictions. CDN outputs are retained for 72 hours.