Anthropic outage history

Anthropic Claude users attempting to create or manage subscriptions via Google Play on Android devices faced a brief disruption. The Claude status page noted a Google Play subscriptions issue, with Google confirming the problem and Claude awaiting further updates. The disruption began at 4:18 PM UTC and was resolved by 4:53 PM UTC.

The scope appeared limited to Google Play-driven subscription flows; no broader Claude services were indicated as affected. The disruption lasted roughly 35 minutes.

An outage affected Claude Mythos 5.1 and Claude Fable 5.1, with intermittent errors observed across multiple Claude components including Claude Cowork, claude.ai, Claude API (api.anthropic.com), and Claude Code. Users attempting to access these models encountered failed or degraded responses beginning at Sep 15, 2026 9:53 AM UTC. The issue persisted as error spikes were reported across requests until a fix was deployed around Sep 15, 2026 10:53 AM UTC, after which monitoring confirmed improvement. By Sep 15, 2026 11:18 AM UTC, the service was returning to normal and status returned to up.

An outage began on Sep 11, 2026 2:23 PM UTC affecting Claude Mythos 5.1 and Claude Fable 5.1 with elevated errors on requests to those models. The incident also touched Claude.ai, the Claude API (api.anthropic.com), and Claude Code, causing user-visible errors for those features. A fix was deployed and recovery monitoring started by Sep 11, 2026 2:33 PM UTC, with services gradually returning to normal for the affected components. Separately, Claude Cowork on Windows showed degraded functionality due to a Windows update that prevented local commands from running, while chat and file reading remained available; Microsoft was actively working on a fix. The Claude Cowork issue moved back to normal operation by Sep 14, 2026 7:18 PM UTC.

Users experienced degraded performance on the Claude API with slower response times for traffic entering through the US Midwest. Some requests timed out, affecting automated workflows and chat interactions that rely on api.anthropic.com. The degraded API performance began at Sep 10, 2026 9:48 PM UTC and continued into the following day. Later that night, Claude Cowork on Windows showed degraded functionality as a Windows update prevented local commands from reaching the workspace drive.

On Sep 11, 2026 2:23 PM UTC, elevated errors were observed for Claude Mythos 5.1 and Claude Fable 5.1. A fix was deployed by Sep 11, 2026 2:33 PM UTC and the team began monitoring recovery. The affected areas included Claude AI services, the Claude API (api.anthropic.com), and Claude Code.

Degraded Windows Cowork functionality persisted at times through Sep 14, 2026 7:18 PM UTC when Claude Cowork moved back to up. User reports were limited, including one from Orlando noting that Claude was not working around 4:39 PM UTC. By Sep 14, 2026 7:18 PM UTC the service had largely recovered, with Claude Cowork restored to normal operation.

Users of Anthropic Claude experienced degraded functionality on Claude Cowork for Windows beginning 3:58 PM UTC after a Windows update released September 8. The issue prevented Claude Cowork from running local commands because the workspace could not reach the computer’s drive. For most users, chat and reading or editing files still worked. Microsoft has developed a fix and is working to release it; there was no in app workaround available.

The incident extended beyond Claude Cowork, with elevated latency and some timeouts on the Claude API for traffic entering through the US Midwest. Elevated errors affected Claude Mythos 5.1 and Claude Fable 5.1 around 2:23 PM UTC. A fix for the affected services was deployed and monitoring followed around 2:33 PM UTC. By 2:38 PM UTC, the Windows de degraded functionality persisted, while Claude API components were moving toward normal operation. Overall, the impact spanned multiple Claude products with shared root causes tied to the Windows update and related service adjustments.

Recovery progressed over the following days, and Claude Cowork returned to an up state by 7:18 PM UTC. The Windows specific issue was resolved and other Claude services resumed standard performance. The incident underscored the complexity of cross component dependencies during the episode and the need for coordinated fixes from both Microsoft and Anthropic.

Users globally experienced degraded Claude service starting at 1:30 PM UTC as elevated errors affected Claude Mythos 5.1, Claude Fable 5.1, and Claude Opus 5 requests. The outage impacted multiple components including Claude Cowork, claude.ai, Claude API (api.anthropic.com), and Claude Code, with responses failing to load and API errors reported across the platform. The disruption lasted for several hours and affected typical model usage and integrations.

Status updates from Anthropic confirmed the investigation and later a fix. An exhaustive list of affected models included Mythos/Fable 5.1, Mythos/Fable 5, Opus 5, Opus 4.8, and Opus 4.6. A fix was deployed by 4:08 PM UTC and monitoring continued; by 4:28 PM UTC all listed components had recovered to normal.

Anthropic Claude Sonnet 5 briefly experienced elevated errors that disrupted requests across several endpoints. The incident began at Sep 3, 2026 12:38 PM UTC and was resolved by Sep 3, 2026 12:48 PM UTC. During the event, Claude Cowork, claude.ai, Claude API (api.anthropic.com), and Claude Code were listed as degraded before all services returned to normal. The status updates described elevated errors while users were attempting to interact with the Sonnet 5 model, with the affected components moving from up to warn and then back to up.

Users attempting to access Claude Sonnet 5 began seeing elevated errors at 9:23 PM UTC. The impact extended to multiple Claude components, including Claude Cowork, claude.ai, Claude API (api.anthropic.com), and Claude Code, resulting in failed requests and degraded interactions for Sonnet 5 workloads. The disruption lasted about 25 minutes, with restoration confirmed by 9:48 PM UTC.

Anthropic engineers investigated the elevated errors and worked to restore traffic. By the time of the resolution, all affected components had returned to normal operation and Claude Sonnet 5 services were available again for users.

Anthropic Claude experienced a credit-related issue that delayed the availability of newly purchased credits, causing some API requests to fail with credit balance issues. The impact affected purchases made between 12:10 PM UTC and 9:35 PM UTC, potentially hindering access to Claude API features that rely on credits.

The Claude Console indicated a degradation in performance, updating to a warning at 11:28 PM UTC as engineers investigated. The problem was resolved and the service returned to normal by 1:28 AM UTC, with newly purchased credits expected to be available again and remaining impact being addressed.

Users of Anthropic Claude experienced degraded performance affecting platform.claude.com and Claude for Microsoft Office 365 starting at 5:08 PM UTC. The issue impacted the Claude Console on platform.claude.com and the claude.ai service (including docs.claude.com), causing slower or intermittent responses for users. By 5:53 PM UTC status updates noted recovery of success rates and that core inference and the API remained unaffected, with full restoration by 6:08 PM UTC. The incident did not constitute a full outage, and services returned to normal after about an hour.