Overbilling for OpenAI-hosted containers in the Agent API

What to do

• Check whether your product depends on OpenAI. Check your stack

  Paste a package.json, lockfile, Terraform or a list of domains at down.guide/stack to see every provider you depend on.

• Confirm your retries, queues and scheduled jobs caught up after the 9h 23m it lasted.

  Anything that failed between 2026-09-18 22:29 UTC and 2026-09-19 07:52 UTC may need a replay.

• Subscribe to OpenAI's feed to hear about the next one first. RSS feed

  https://down.guide/feed?provider=openai works in Slack (/feed subscribe), Discord feed bots and any RSS reader.

Updates from OpenAI 5

1. investigatingSep 18, 2026, 10:29 PM UTC

   We’re investigating an issue causing higher-than-expected charges for OpenAI-hosted containers in the Agent API. We’re working on a fix and preparing refunds for affected customers. We’ll share further updates as we make progress.

   provider

2. investigatingSep 18, 2026, 11:22 PM UTC

   We’re continuing to work on a fix for the container overbilling issue. We’re also reviewing affected usage to identify impacted customers and calculate refunds.

   provider

3. identifiedSep 19, 2026, 1:08 AM UTC

   We are still implementing the mitigation.

   provider

4. monitoringSep 19, 2026, 5:00 AM UTC

   We have applied the mitigation, and any new sessions would not run into this issue. We are monitoring the recovery.

   provider

5. resolvedSep 19, 2026, 7:52 AM UTC

   All impacted services have now fully recovered.

   provider

Corrections

None so far.

Something wrong on this page? Report an error or send us a note. Every fix is listed here.