AI incidents: OpenAI cost-data delay, image API recovery and Kits response — 15 October 2026

Publisher disclosure: MUVO offers competing music, remix, and video tools. This report summarizes public sources and provider documentation; it is not an independent product test or a claim that MUVO is free of similar issues. Read our reporting method.

Incident summary and status at the research cutoff

OpenAI’s cost-data reporting delay remains unresolved in its official notice; a separate problem affecting two image API models has recovered. This edition also records a Kits support response and two dated Kling accounts about output quality and credit use.

The scan covers 12 October at 13:26 through 15 October at 12:14 Europe/Istanbul (UTC+3). Dates below identify actual source events or customer experiences, not necessarily events that began on 15 October.

OpenAI: organization cost-data delay remains identified

OpenAI says organization cost data is arriving late through its Costs API. The incident is filed under the Compliance API component. Its notice explicitly says API and ChatGPT usage are unaffected and other reporting channels remain available. This does not establish a generation outage or an incorrect charge. Official incident.

All times here are Istanbul time on 14 October:

• 20:00: First public acknowledgement, already marked identified.
• 20:02: Follow-up identifies the affected status component.
• 23:00: OpenAI says a fix is undergoing testing; status remains identified.

Impact start, detailed cause, recovery time and regions are unspecified; duration cannot be established. The ledger preserves UTC timestamps.

For cost-reporting workflows, consult the provider’s available alternative channels and follow the incident for restored data freshness.

Update, 16 October: At 01:36 Istanbul, OpenAI reported a deployed fix and moved to monitoring. Cost data remains delayed while its backlog is processed; full recovery is not confirmed. Provider update. The original cutoff above is preserved; see the 16 October follow-up.

OpenAI: two GPT Image 2.5 API models recover

OpenAI marked limited availability for gpt-image-2.5-flare and gpt-image-2.5-sunburst resolved on 14 October. The notice identifies API image generation, without extending the event to every image model, ChatGPT or Sora. Official image API incident.

The public sequence, in Istanbul time, is 13:18 investigation, 13:40 mitigation applied and monitoring, and 13:58 provider-reported recovery. These are update times, not a demonstrated start and end of customer impact. The source does not identify a root cause, region or specific workaround, and does not establish that every previously failed request completed.

Kits AI: support responds to the download-access review

Kits has invited the customer behind the 3 October download-access account to contact support. We observed the reply on 15 October. Its displayed time was relative, so its exact publication timestamp remains unknown. The response provides an investigation route but does not confirm restored downloads, an adjusted minute balance or a refund. Original listing and reply.

This is a continuation of the existing Kits download-minutes report, whose original date and URL are retained. It is not a new platform outage. The current Kits pricing FAQ still separates conversion allowances from download allowances; it cannot establish the outcome of this customer’s account.

Kling: two users report output problems consuming credits

Two separate accounts describe results that failed to match their prompts while consuming credits. One has a 12 October experience date and describes skipped or inaccurate renders. The other has a 15 October experience date and describes unusable output from paid tokens and an unanswered support email. Kling customer-review listing.

These are customer-reported experiences, not a confirmed outage, billing defect or measured quality regression. Their precise publication timestamps were not available in the rendered listing, and the individual review permalinks could not be fetched. Neither account identifies a model version or supplies reproducible settings. No case-specific company response was visible for those accounts. The sample cannot establish frequency or whether the two experiences share a technical cause.

Separate accounts about waiting times or a busy free tier are not evidence of the same issue. We have not merged them into a claimed widespread disruption.

Sources

• OpenAI: Costs API reporting delay and image API recovery, both dated 14 October.
• Kits: customer accounts and provider responses, plus current pricing guidance.
• Kling: review listing; experience dates 12 and 15 October. Individual permalinks are retained in the ledger despite retrieval failures.

Anthropic’s reporting recovery and ElevenLabs’ earlier recoveries were already covered in the 14 October edition. Rechecking them did not establish another new event. Google’s Cloud history did not establish a new AI incident; its dedicated Gemini API status page was inaccessible. Other inaccessible sources are recorded as coverage gaps, not evidence that services were healthy.

We have not inspected private customer accounts or independently reproduced these issues. This is a daily scan, not continuous monitoring. Product releases are covered separately in today’s AI news and updates.

Update timeline

• 15 October 2026, 12:14 Istanbul: Research cutoff for this edition. Two official OpenAI events, a Kits response and qualified Kling accounts consolidated. Publication followed later on 15 October.

• 16 October 2026, 12:00 Istanbul: Added the later cost-data monitoring update. Original publication date, URL and cutoff retained; no claim of complete recovery.