Superpower Daily
GPT-6 Astra Hits 99.9% on ARC-AGI-3, but Scores 62.7% in a Shar...Crusoe Signs Reported $13 Billion, Five-Year AI Cloud Deal With...Microsoft Releases MAI-Transcribe-2 at 10 Cents an HourMeta Lawsuit Adds Bystanders Who Say AI Glasses Recorded Them W...Figure Signs $3.5B Nscale Deal for Up to 100,000 GPUs for Human...xAI Explains How Grok Bot Is Built to Keep Agents Working Betwe...GPT-6 Astra Hits 99.9% on ARC-AGI-3, but Scores 62.7% in a Shar...Crusoe Signs Reported $13 Billion, Five-Year AI Cloud Deal With...Microsoft Releases MAI-Transcribe-2 at 10 Cents an HourMeta Lawsuit Adds Bystanders Who Say AI Glasses Recorded Them W...Figure Signs $3.5B Nscale Deal for Up to 100,000 GPUs for Human...xAI Explains How Grok Bot Is Built to Keep Agents Working Betwe...
Culturepublished

Four Major AI Services Suffer Rare, Overlapping Outages

OpenAI and Anthropic restored their reported incidents within hours, while Grok still showed a service error at publication. The near-simultaneous failures also left an important question unanswered: whether the outages had any common cause.

By 2 min read
Four Major AI Services Suffer Rare, Overlapping Outages
Four Major AI Services Suffer Rare, Overlapping Outages

Listen to this story

The audio brief

About 1:32
0:001:32
Read transcript
Four of the most prominent AI services—ChatGPT, Claude, Gemini, and Grok—reported disruptions in the same Thursday-morning window, an unusually tight cluster. Anthropic reported elevated errors for Claude Mythos 5.1, Fable 5.1, and Opus 5 starting at 9:23 a.m. Eastern. It said the cause was identified about 15 minutes later, and the incident was resolved at 12:16. A shorter error period then affected Claude Sonnet 5. OpenAI reported degraded performance across ChatGPT and Codex at 10:43 a.m. It deployed a mitigation a little over half an hour later, and declared the issue resolved at 12:55. Grok was the outlier: user reports peaked at 1,365 on Downdetector by 9:45 a.m., and xAI was still showing users an error later in the day. Gemini is the least certain case. Google did not acknowledge an outage, but Downdetector reports climbed sharply, while StatusGator recorded what it called a likely disruption lasting from about 10:45 to 11:15. The timing is striking, but it does not prove a shared failure. Amazon Web Services, Microsoft Azure, and Cloudflare reported no major incident, and no common cause had been identified at publication. The practical takeaway is more immediate: teams relying on chatbot interfaces need fallback tools. The open question is whether these were truly independent failures—or a common dependency that has not yet been found.

Story brief

3 key points

A rare cluster of reported disruptions hit ChatGPT, Claude, Gemini and Grok on the same Thursday morning, but the evidence does not establish a shared infrastructure failure. Anthropic and OpenAI resolved confirmed incidents within roughly three hours, while Grok remained impaired later; Gemini’s outage was inferred from third-party monitoring rather than acknowledged by Google. The episode highlights growing...

  1. 01

    Anthropic logged Claude errors from 9:23 a.m. to 12:16 p.m. Eastern, including Mythos 5.1, Fable 5.1 and Opus 5.

  2. 02

    OpenAI reported ChatGPT and Codex degradation at 10:43 a.m.; mitigation followed, with resolution declared at 12:55 p.m.

  3. 03

    Grok reports peaked at 1,365 on Downdetector by 9:45 a.m.; xAI still showed a user-facing error later.

ChatGPT, Claude, Gemini and Grok all experienced reported service disruptions during the same Thursday-morning window, an unusually overlapping failure across four prominent AI services. OpenAI and Anthropic later marked their incidents resolved, but Grok was still displaying a user-facing error when Ars Technica published its account.

Anthropic was first to post a partial-outage notice, reporting elevated request errors for Claude Mythos 5.1, Claude Fable 5.1 and Claude Opus 5 at 9:23 a.m. Eastern. It said it identified the cause about 15 minutes later, deployed a fix and resolved the incident at 12:16 p.m.; a separate brief error period affected Claude Sonnet 5 shortly after noon.

OpenAI reported degraded performance from elevated errors across ChatGPT and Codex at 10:43 a.m. It deployed a mitigation a little more than half an hour later and marked the incident resolved at 12:55 p.m.

Grok remained affected later in the day. Its service presented users with an error saying it was experiencing issues and that xAI was working to restore service. User-submitted Downdetector reports for Grok rose from fewer than 10 shortly before 9 a.m. to 1,365 by 9:45 a.m., before falling later.

Gemini is the least conclusive case. Google had not publicly acknowledged a problem, but Downdetector reports rose from roughly 23 to 412, and StatusGator recorded what it called a likely Gemini API outage between 10:45 and 11:15 a.m. The signals support a reported disruption, not a confirmed account from Google of its scope or cause.

The timing was striking, but it does not by itself show a common failure. Amazon Web Services, Microsoft Azure and Cloudflare had not reported major issues at publication, although user issue reports for all three rose somewhat that morning. Anthropic and OpenAI posted incident responses, but no shared cause had been identified.

The disruption also drew jokes about how much work and attention now run through chatbot interfaces. That reaction is cultural rather than evidence of operational impact, but it accompanied an incident that was rare by the providers’ published reliability figures: Claude reported 99.4% uptime over the preceding 90 days, while OpenAI reported 99.63% for ChatGPT and 100% for ChatGPT Codex over the same period.

Sources

  1. arstechnica.comFour major AI models suffer rare overlapping downtime
  2. futurism.comWorld Plunged Into Chaos as ChatGPT, Claude, and Grok Suddenly Go Down Simultaneously: "Finally I Can See the Sun!"

Related coverage

Continue reading