gpt-image-2-all occasionally hits 300s timeouts · related to content-safety blocks
Our investigation shows that the long-running gpt-image-2-all requests are mostly ones where a content-safety check failed and triggered account-pool retries — the stacked retries stretch the total duration. It is not a performance degradation of the channel itself.
Usage advice:
- Keep prompts and input images compliant and reasonable, to reduce safety-triggered retries at the source
- A client
timeoutof 300s is fine, and shortening it somewhat is also OK to avoid long waits
gpt-image-2-all was billed and the duration reached 300s, APIYI will reissue the credits in one batch, expected on Saturday, 2026/7/11 (UTC+8).
Thanks for your understanding — we keep operating and improving.
← Back to Live Updates · 📚 Monthly Archive