AI service incidents · Daily incidents
AI incidents: Claude model errors and OpenAI Costs API recovery — 16 November 2026
Anthropic deployed a fix for Claude model errors. OpenAI restored cost reporting, and Fireworks resolved a DeepSeek inference degradation.
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.
Anthropic deployed a fix for Claude model errors
Anthropic opened an incident for elevated errors on Claude Sonnet 5.5 and Claude Fable 5.2 at 10:49 Istanbul on 16 November. At 11:02, it said a fix had been deployed for Sonnet 5.5 and other Claude models and moved the incident to monitoring. Its notice lists claude.ai, the Claude API, Claude Code and Claude Cowork as affected. Anthropic had published no exact impact start, error rate or resolution at cutoff. Claude model incident.
The separate issue with Sign in with Apple in the Claude iOS app was still under investigation. It mainly affected new accounts. Anthropic continued to recommend using Google or email to sign up. Claude iOS sign-up incident.
Anthropic later confirmed both recoveries
Anthropic posted the model incident resolution at 12:04 Istanbul on 16 November and dated the impact to 10:32–11:45, a 73-minute interval. It later closed the iOS sign-up incident at 22:52, dating the impact from 03:29 on 14 November to 21:31 on 16 November. The issue mainly affected new accounts using Apple sign-up. See the 17 November follow-up for the related timeline and the separate Claude Meetings beta disruption.
OpenAI marked cost reporting restored
OpenAI reported normal operation at 03:17 Istanbul on 16 November after delays publishing organization cost data through the Costs API. It first disclosed the issue at 19:10 on 13 November, reported applying a mitigation at 21:56, and still expected data delays in its 08:31 update on 14 November. OpenAI said API and ChatGPT usage were unaffected. The notice supplies a component timeline but gives no separate start for the underlying delay or a time when individual organizations’ data caught up. The 15 November edition now includes the resolution. OpenAI Costs API incident.
Fireworks restored its DeepSeek V4.1 Flash 1017 endpoint
Fireworks reported degraded performance on its hosted DeepSeek V4.1 Flash 1017 endpoint from 22:53 to 23:10 Istanbul on 15 November. The interval was about 17 minutes, based on the provider’s recorded component start and end. Fireworks reported that the endpoint was operational again. The notice gives no cause, error-rate figure or regional breakdown. Fireworks incident.
Sources
The Anthropic incident archive, OpenAI incident history and Fireworks incident history support this edition. All displayed times use Istanbul time (UTC+3). Public quality complaints and counterreports did not establish a new, dated service incident. Reporting methodology.
Update timeline
- 16 November 2026, 11:47 Istanbul: Research cutoff. Included the Claude model fix, OpenAI cost reporting recovery and Fireworks endpoint restoration.
- 17 November 2026: Added Anthropic’s confirmed impact windows and resolution notices for model errors and iOS sign-up.