Claude outage history

Claude experienced a brief outage caused by a server overload that temporarily degraded the service. Users began encountering 529 Overloaded errors and general API unavailability starting at 11:00 AM UTC, with the issue appearing to resolve by 11:01 AM UTC. Reports came from multiple locations, including Mumbai and Rotterdam, indicating a short, cross-regional impact rather than a localized problem.

The Claude family of services experienced a global outage starting around Sep 29, 2026 2:00 PM UTC, affecting Claude AI (including desktop and mobile apps), Claude Code, Claude Cowork, and the Claude API. Users faced elevated errors, failed requests, sign‑in prompts, and conversations that would not load or save reliably. The incident unfolded in waves: an initial outage with degraded request handling, followed by a second set of authentication and session issues that prevented signing in, starting new chats, and uploading files.

Claude's AI suite encountered an outage that caused elevated errors and partial unavailability across Mythos 5.1, Fable 5.1, and Opus 5. The incident began around 1:00 AM UTC after reports of server overload, with users experiencing unresponsive chats and failed requests.

Status page updates showed rapid progression. The team identified the cause and worked to resolve the issues; by 1:38 AM UTC Mythos and Fable had resumed normal operation, while Opus continued to experience errors. Full restoration across Claude endpoints was achieved by 2:13 AM UTC.

StatusGator’s Early Warning Signal captured the issue at Sep 22, 2026 12:59 AM UTC, about 45 seconds before Claude’s official acknowledgment at 1:00 AM UTC. The outage appears broad in scope, with a high volume of user reports indicating global impact rather than a regional problem.

A brief disruption affected Claude users attempting to create new subscriptions via Google Play on Android devices. The incident began at 4:18 PM UTC and was resolved by 4:53 PM UTC. Claude's status page indicated issues with Google Play subscriptions and noted that Google had confirmed an issue and that updates would follow. The disruption was limited to the Google Play subscription flow and did not indicate a broader outage across Claude services; users likely encountered errors or stalled subscription attempts during the window as the team coordinated with Google for a fix.

Intermittent error spikes affected Claude Mythos 5.1 and Claude Fable 5.1, impacting requests routed through Claude Code, Claude Cowork, claude.ai, and the Claude API (api.anthropic.com).

The disruption began at 9:53 AM UTC with elevated errors and degraded responses. A fix was deployed by 10:53 AM UTC and monitoring continued to ensure stability. By 11:18 AM UTC the affected components had returned to normal operation.

Users relying on Claude Mythos 5.1 and Claude Fable 5.1 began seeing elevated errors starting 2:23 PM UTC, with impact spreading to Claude API endpoints (api.anthropic.com), Claude Code, and related tooling. The disruption caused failed requests and degraded experiences for developers and workflows that depended on Claude. A fix was deployed and core Claude services returned to normal by 2:33 PM UTC, with monitoring ongoing. However, a Windows-specific degradation affecting Claude Cowork on Windows emerged around 2:38 PM UTC, leaving local commands temporarily unavailable for some users while chat and file reading remained functional for many.

Claude users experienced degraded performance on the Claude API beginning around 9:48 PM UTC, with slower response times and occasional timeouts for traffic entering through the US Midwest. The impact extended to Claude Mythos 5.1 and Claude Fable 5.1, and in some cases Claude Code, with elevated latency and errors reported by users worldwide. A separate Windows-specific degradation affected Claude Cowork on Windows after a September 8 update, leaving the workspace unable to access the local drive; most chat and file operations remained possible, but there was no in-app workaround while Microsoft worked on a fix.

Recovery progressed over the following days as a fix was deployed around 2:33 PM UTC and monitoring continued. By 7:18 PM UTC Claude Cowork had returned to up, and Claude API and related services were returning to stable operation. The footprint appeared global, with user reports spanning multiple regions across the Americas, Europe, and beyond, indicating a broad impact before the service fully recovered.

Users faced degraded Claude Cowork functionality on Windows beginning at 3:58 PM UTC after a Windows update released on September 8. While chat, reading, and editing generally worked, local commands could not run because the workspace could not reach the PC's drive. Microsoft has a fix in progress, and there was no in-app workaround yet.

Around 9:48 PM UTC, some customers began experiencing slower than normal responses on the Claude API, particularly for traffic entering through the US Midwest. This manifested as elevated latency and some request timeouts, with additional latency reports at 10:18 PM UTC and 11:28 PM UTC as the issue persisted.

An outage began around 1:30 PM UTC after elevated errors affected Claude Mythos 5.1, Claude Fable 5.1, and Claude Opus 5. Users attempting to use Claude API (api.anthropic.com), Claude Code, Claude Cowork, or claude.ai encountered failed requests and timeouts as the models degraded. StatusGator logged an Early Warning Signal at Sep 3, 2026 1:25 PM UTC, about five minutes before Claude acknowledged the problem, indicating early user-facing signs of the outage. The disruption persisted for most of the afternoon, affecting multiple models and services across the Claude suite.

Users attempting to use Claude Sonnet 5 encountered elevated errors across Claude API (api.anthropic.com), Claude Code, Claude Cowork, and claude.ai. The incident began at Sep 3, 2026 12:43 PM UTC with elevated errors on requests to Claude Sonnet 5 and a notification that engineers were investigating. The disruption lasted about five minutes and was resolved by Sep 3, 2026 12:48 PM UTC as all affected components returned to normal and service health was restored.