AI services & developer workflows · Daily incidents

AI incidents: OpenAI and Copilot recoveries, Fireworks Kimi degradation — 10 November 2026

OpenAI resolved API, ChatGPT conversation and Costs API export issues. Copilot and Cursor recovered. The separate Fireworks Kimi K3 notice is now unavailable; its current status is unverified.

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.

OpenAI recovered; Fireworks Kimi status needs a follow-up

OpenAI resolved API errors, ChatGPT conversation errors and an issue with exported cost data. GitHub Copilot and Cursor reported separate recoveries. Fireworks’ Kimi K3 US incident is resolved, but its separate Kimi K3 notice was investigating at the original cutoff. Its current status is unverified because the public notice is now unavailable. These notices describe different products and components; they do not establish one continuous, shared outage.

Coverage runs from 7 November at 11:52 through 10 November at 12:05 Europe/Istanbul (UTC+3), including a 48-hour overlap from the previous scan. Times below use Istanbul time. A provider’s resolution does not establish that every previously failed request or job subsequently completed.

Service / issue Latest status at cutoff Useful notice times
OpenAI API errors Resolved 10 Nov: identified 02:08; resolved 02:50
ChatGPT Conversations Resolved 9 Nov: mitigation announced 21:16; resolved 22:08
OpenAI Costs API exports Resolved 10 Nov: acknowledged 07:42; resolved 10:02
GitHub Copilot model availability Resolved 10 Nov: investigation 02:16; resolved 03:05
Cursor OpenAI model requests Resolved 10 Nov: investigation 02:10; resolved 02:44
Fireworks Kimi K3 US Resolved 10 Nov: investigation 08:50; resolved 09:56
Fireworks Kimi K3 Investigating 10 Nov: first notice 09:01

11 November update: The separate Kimi K3 permalink now returns no incident details and is absent from the retrieved history. Its current status is unverified; the investigating entry above describes the original 10 November cutoff, not a confirmed continuing failure. OpenAI also resolved its older Mastercard incident at 19:06 on 10 November. See the follow-up roundup.

These are notice timestamps, not independently measured outage boundaries. Original sources and scope details follow.

OpenAI: API errors, conversations and cost exports were separate issues

API errors resolved across named components

OpenAI marked its API error incident resolved at 02:50 on 10 November. The earlier 02:08 notice said it had identified elevated errors and was implementing mitigation. The incident tagged Chat Completions, Responses, Agents, Decisions, Fine-tuning, Embeddings, Images, Batch, Audio, Moderations, Realtime, Files, Login and Live. It did not say that every request failed or identify particular regions. OpenAI API incident.

Its component records span 02:08:12–02:50:46, about 43 minutes. This is the provider-recorded degraded interval, coinciding with its acknowledgment and resolution; the earliest user impact is unknown. No technical cause was disclosed.

ChatGPT conversation mitigation preceded the first notice

The ChatGPT Conversations component recovered at 22:08 on 9 November, according to OpenAI. Its first published update at 21:16 already said mitigation had been applied. That update therefore cannot establish when the disruption began. The notice gives no affected plan, model, region or technical cause. ChatGPT conversation incident.

Costs API exports contained incorrect data

OpenAI acknowledged incorrect cost data in Costs API exports published between 6 and 9 November. At 07:42 on 10 November, it said it was correcting affected records; at 10:02, it marked the incident resolved. The notice identifies exported data, without establishing incorrect invoice charges or specifying which accounts were affected. Exact impact hours and a technical cause were not supplied. Costs API export incident.

Practical analysis: If an internal usage report used those exports, compare it with newly retrieved records before treating its totals as final. This is a data-validation suggestion, not an official refund or billing remedy.

Copilot and Cursor recovered from model-request problems

GitHub resolved degraded availability for GPT-5.3 Codex, GPT-5.6 Luna and GPT-5.6 Sol in Copilot products and IDE surfaces at 03:05 on 10 November. It first acknowledged the issue at 02:16, attributed it to an upstream model provider, and recommended another model or Auto during the disruption. Model-provider mitigation was reported at 02:46, followed by a model-specific monitoring update at 02:59. A detailed root-cause analysis was promised but absent at cutoff. An earlier generic “operating normally” update at 02:22 was not the incident’s final resolution. GitHub Copilot notice.

Cursor separately reported failed or incomplete OpenAI model responses. Its notice covered Cursor IDE, CLI, Cloud Agents, Automations and third-party models. Investigation began at 02:10; it reported recovery and monitoring at 02:41, then resolution at 02:44. Other models were unaffected and were its suggested alternative. Cursor notice on the xAI status site, original incident feed.

The Cursor notice does not establish a Grok-wide disruption. The overlapping OpenAI, Copilot and Cursor timelines alone are not proof that all three notices share the same technical cause. User-impact start times were not specified for the two downstream services.

Fireworks: Kimi K3 US recovered; separate Kimi status now unverified

Fireworks resolved its Kimi K3 US degradation at 09:56 on 10 November, after first reporting impacted performance at 08:50. The separate Kimi K3 notice, opened at 09:01, still said investigating at cutoff. Neither notice explained a cause or supplied a workaround. The distinction between the two named components matters: recovery of the US component does not resolve the second notice. Kimi K3 US, Kimi K3.

Follow-up, 11 November: The separate Kimi K3 page no longer exposes a timeline. No resolution time was verified, and disappearance alone is not recovery evidence. The US incident remains independently documented.

These are Fireworks service notices; they do not establish an outage across every operator offering a Kimi model. Actual impact start times remain unknown.

GitHub workflow disruptions and Bolt prompting also resolved

On 9 November, GitHub reported delayed Actions starts and webhook deliveries, first acknowledged at 13:00. Initial mitigation at 13:47 did not end the notice: Webhooks remained degraded, entered monitoring at 14:45, and the incident resolved at 14:50. Actions and Webhooks notice.

A separate GitHub incident began with an Actions notice at 21:12, later identifying elevated Actions and Checks API errors, plus Webhooks and Pages degradation. Mitigation was announced at 21:43, followed by resolution at 21:59. Both GitHub notices promised later root-cause analyses; neither supplied defensible total impact boundaries. These developer-workflow events are distinct from Copilot’s model incident. Actions and Checks API notice.

Bolt’s Prompting feature was marked degraded at 12:50 on 9 November and resolved at 12:57. The short notice did not describe a technical cause, broader product impact or a workaround. Bolt prompting notice.

Update: yesterday’s GitHub anonymous-request incident recovered

GitHub resolved the recurring anonymous-request timeouts at 12:09 on 9 November, after yesterday’s 11:52 cutoff. Monitoring resumed at 12:05. Its latest explanation said error rates remained low and additional protections had been prepared; logged-in users and authenticated requests were unaffected. We have updated the original report while preserving its publication date and recurrence timeline. This is a continuation, not a new outage. Original incident.

OpenAI’s U.S. business Mastercard notice and Fireworks’ deployment, training and batch-inference notice retained their earlier monitoring updates at the original cutoff. Update, 11 November: OpenAI subsequently resolved its payment incident at 19:06 on 10 November; Fireworks’ older restoration/monitoring notice remains unchanged.

Sources

Original notices are linked beside the findings. We also rechecked Anthropic’s history, Google Cloud, Google Workspace, ElevenLabs’ US and EU histories, creative-tool sources and accessible original reviews. No additional qualifying incident was established for Anthropic, Google or ElevenLabs. A follow-up check of Google AI Studio’s rendered history found no listed incident newer than 30 October. Blocked reviews and incomplete archives leave further gaps. This is not a claim of universal service health. See our reporting method.

Update timeline