Status›History›inc-2026-04-22

Brief AI generation latency

April 22, 2026 · 23 minutes · minor
  1. Resolved
    April 22, 14:18 UTC
    AI generation latency back to baseline. Root cause: an upstream rate limit at our AI provider; we've added local caching for prompts to reduce upstream calls during similar conditions.
  2. Identified
    April 22, 14:02 UTC
    Identified upstream rate limit as the cause. Engaging the AI provider's support team and routing prompts through a fallback model.
  3. Investigating
    April 22, 13:55 UTC
    Investigating elevated latency on /api/ai/* endpoints. Image-to-system requests are queueing.