Anthropic’s Claude suffered three separate outages overnight, with the longest disrupting the service for 35 minutes, according to the company’s official status page. The main incident ran from 10:00 to 10:35 PKT on Saturday, October 24, after two earlier incidents in the small hours, and Anthropic says all services are now back to normal. The first incident began at 04:28 PKT, affecting Claude Fable 5.1, Sonnet 5.5 and Haiku 5.5, followed by a brief Opus 5.5 incident at 05:23 PKT. Anthropic said it identified the cause in each case but has not published what went wrong, leaving users with a fix and no explanation so far.
Three Incidents In About Six Hours
The first incident began at 23:28 UTC on October 23, which was 04:28 PKT on October 24. Anthropic reported elevated errors on requests to Claude Fable 5.1, Sonnet 5.5 and Haiku 5.5, as well as fast-mode requests to Opus 5.5. The company said it was investigating within minutes, identified the cause at 23:39 UTC, and saw success rates return to normal before midnight UTC. The status page records the impact as lasting from 23:28 to 23:59 UTC, about half an hour.
The second incident was shorter. At 00:23 UTC on October 24, or 05:23 PKT, Anthropic reported elevated errors on requests to Claude Opus 5.5. The impact lasted until 00:34 UTC, about 11 minutes, and the incident was marked resolved at 00:37 UTC.
The third and largest incident began at 05:00 UTC on October 24, or 10:00 PKT. Anthropic reported elevated errors on requests to other Claude models, later describing it as an issue affecting Claude services. According to Paktechstan, which reported the overnight outages on Saturday afternoon, the disruption affected claude.ai, the Claude API, Claude Code and the Claude Cowork agent tool. Anthropic identified the cause at 05:24 UTC, deployed a fix at 05:41 UTC, and marked the incident resolved at 05:53 UTC. The recorded impact ran from 05:00 to 05:35 UTC, or 35 minutes.
What Anthropic Has And Has Not Said
Anthropic’s status page now lists every Claude component as Operational, including claude.ai, the Claude Console, the Claude API, Claude Code, Claude Cowork and Claude for Government. That is the full extent of the company’s public statement on the night: each incident entry records that a cause was identified and a fix was deployed, but none of the entries says what the cause was.
Paktechstan, in its report published at 13:47 PKT on Saturday, confirmed the same point, noting that Anthropic has not published a detailed explanation for any of the three incidents. The status updates confirm only that the issues were identified, fixed and resolved, with success rates back to normal across the affected models and services.
No further incident has been recorded since the third one was resolved. The status page carries no warning of continued instability and no estimate of whether the pattern will repeat.
Why It Matters
Claude’s overnight wobble matters because the service is no longer only a chatbot people open in a browser. It powers coding assistants, customer-support tools, automated workflows and products that companies ship to their own customers, so when the model stops answering, those downstream tools stall too. The first sign of trouble for an end user is often a confused customer, not an error message.
What stands out about this night is the repetition rather than the length. A single 35-minute outage is a routine event for a large AI platform handling spiky demand from millions of users. Three separate incidents in about six hours, touching different models and then the wider service, is a harder pattern to dismiss, and teams running Claude inside production systems will be watching whether Saturday morning’s fix holds. The practical lesson is an old one for anyone building on a hosted model: build in retries, fallbacks and clear error messages, because even the most reliable services go quiet sometimes.
Background
Claude is Anthropic’s family of AI assistants, available through claude.ai, an API, the Claude Code coding tool and the Cowork agent tool, and Anthropic publishes a public status page tracking each of those services. What Claude is and what it can do has changed quickly this year, with auto mode becoming the default for Pro, Max and Team sessions from next week. The overnight incidents were not the week’s first: the same status page records a brief incident on October 22 affecting claude.ai and Claude Cowork.
Source: Anthropic Claude Status, Paktechstan
0 Comments