generated: '2026-07-20'
method: searched
source: openapi/pruna-ai-openapi.yml
notes: >-
  Rate limits documented in the P-API OpenAPI description and surfaced via HTTP
  429. Standard prediction throughput is credit/quota governed rather than a
  published fixed RPM; trainer models carry an explicit lower limit.
signalled_via: HTTP 429 Too Many Requests
rate_limits:
  - scope: trainer models (p-image-trainer, p-image-edit-trainer)
    limit_count: 5
    window: minute
    description: LoRA trainer endpoints are limited to 5 requests per minute.
  - scope: file uploads (POST /v1/files)
    description: >-
      File uploads are rate limited (HTTP 429). No fixed count published;
      governed alongside account quota.
quota:
  model: credit-based
  description: >-
    Usage is metered against purchased credits (minimum $5 top-up); exceeding
    quota returns a QUOTA_EXCEEDED prediction error.