Anthropic Outage History

An extended outage affected Anthropic Claude, beginning at Feb 25, 2026 2:03 PM UTC with elevated errors on Claude Sonnet 4.6 and Opus 4.6, impacting user requests across Claude platform services. The issue expanded to Claude Desktop on Windows, with startup failures for a subset of users as multiple components were listed as affected (platform.claude.com, claude.ai, Claude Code). By Feb 25, 2026 4:23 PM UTC, Windows desktop failures were actively being worked on and API calls began to show 500s with elevated error rates across several Claude models. Over the next several hours, users experienced degraded performance and outages across multiple parts of the Claude ecosystem.

Anthropic Claude Sonnet experienced elevated errors affecting Claude Sonnet 4.5, with later clarification that Sonnet 4.6 was impacted as well. The outages disrupted access to Claude AI services and related components, including platform.claude.com, claude.ai, and Claude Code, causing users to experience delayed or unresponsive behavior.

The incident began at Feb 24, 2026 12:23 PM UTC when the status page first reported elevated errors. A follow-up update at Feb 24, 2026 1:03 PM UTC confirmed ongoing issues involving the 4.6 model. A final update at Feb 24, 2026 1:28 PM UTC restored all components to normal. The provider issued three status updates during the incident.

User reports from multiple regions, including Paris, Bengaluru, and Zagreb, indicate a broad, global impact. The outage lasted about 1 hour and 5 minutes, with services returning to normal after the final update, affecting three main components: platform.claude.com, claude.ai, and Claude Code.

The Claude service experienced a multi-model outage beginning Feb 18, 2026 3:28 AM UTC and lasting roughly six days, with restoration underway by Feb 24, 2026 11:53 AM UTC. The incident affected Claude Opus 4.6, Claude Sonnet 4.5, Haiku 4.5 and related components, including platform.claude.com, claude.ai, Claude API, and Claude Code. Users faced chat failures, degraded performance, and coding and file-upload issues across multiple models as status transitions moved from elevated errors to outages and back toward normal operation. The outage timeline included frequent status updates and progressive recovery across components, with the last listed recovery occurring on Feb 24, 2026 11:53 AM UTC.

Anthropic Claude Opus 4.6 experienced elevated errors starting at Feb 16, 2026 4:08 PM UTC, affecting platform.claude.com, claude.ai, Claude API (api.anthropic.com), and Claude Code. Users reported issues and degraded access, including a Preston, United Kingdom user seeing a 503 error on the file reader around 4:15 PM UTC.

By Feb 16, 2026 4:58 PM UTC, a fix had been implemented and monitoring continued. All components returned to up by Feb 16, 2026 7:18 PM UTC. The incident involved three official status updates during the period.

The incident involved elevated errors across Anthropic Claude Opus 4.6, impacting Claude AI, the Claude API (api.anthropic.com), and Claude Code. Users may have experienced intermittent failures or degraded functionality during the event. The outage lasted about 35 minutes, beginning at Feb 16, 2026 2:33 PM UTC and ending at Feb 16, 2026 3:08 PM UTC.

The Claude status page documented three updates during the incident: Feb 16, 2026 2:33 PM UTC indicating investigation; Feb 16, 2026 2:43 PM UTC noting a fix and monitoring; and Feb 16, 2026 3:08 PM UTC confirming recovery as components returned to normal.

The Claude Code Execution Tool experienced a brief outage, causing API requests using that tool to fail for users. The incident began at 4:13 PM UTC and was resolved by 4:43 PM UTC, lasting about 30 minutes. The Claude status page described the event as 'Issues with Code Execution Tool' and showed engineers investigating, followed by restoration of service after two official updates.

The outage began at Feb 14, 2026 7:28 AM UTC with elevated errors across Claude API (api.anthropic.com), Claude AI, platform.claude.com, and Claude Code. Users experienced prompts failing to load, chat interactions not responding, and API requests failing. The incident lasted about 1 hour 20 minutes and all services were restored by Feb 14, 2026 8:48 AM UTC.

The Anthropic status page issued two updates during the incident, describing initial investigation into elevated errors and then restoration across the affected components. User reports came from multiple regions, suggesting broad, worldwide impact.

Anthropic Claude experienced a degradation beginning at Feb 13, 2026 5:14 PM UTC that affected four components: platform.claude.com, claude.ai, Claude API (api.anthropic.com), and Claude Code. Users attempting to access Claude features may have encountered errors or degraded performance across these services. The incident lasted until Feb 13, 2026 11:23 PM UTC, for an approximate duration of 6 hours and 9 minutes. Three status updates were issued during the incident, noting elevated errors and subsequent recovery.

Anthropic reported elevated errors on Opus 4.6 affecting four Claude components: platform.claude.com, claude.ai, Claude API (api.anthropic.com), and Claude Code. Users could have encountered API failures and degraded interactions with the Claude console/UI during the incident. The outage began at Feb 12, 2026 2:19 PM UTC and was resolved by Feb 12, 2026 3:19 PM UTC.

A small number of user reports appeared around 3:00 PM UTC, indicating users experienced issues. The provider posted two status updates during the incident, noting elevated errors on Opus 4.6 and the subsequent restoration of normal operation. The incident lasted about one hour before services returned to normal.

Anthropic Claude users faced degraded performance across multiple models and endpoints starting with Claude Sonnet 4.5, where elevated errors were first reported at 1:23 PM UTC. A subsequent issue affected Claude Haiku 4.5 at 2:58 PM UTC, broadening the impact to other Claude components. The problem impacted four main Claude endpoints: platform.claude.com, claude.ai, Claude API (api.anthropic.com), and Claude Code. The incident persisted until 3:48 PM UTC, lasting approximately 2 hours and 25 minutes, during which users experienced partial outages and slower responses.