Claude Outage History

Users experienced degraded performance across Claude Sonnet 4.6 tools as errors surfaced in the Claude API (api.anthropic.com), claude.ai, and Claude Code. The incident began at 4:27 PM UTC with elevated errors, and the Claude status page flagged a degraded state for these components at 4:38 PM UTC. Users reported chats failing to respond or taking longer than usual across many regions, indicating a global impact. The issues were resolved and services returned to normal by 7:48 PM UTC after the status changed to up.

Users of Claude Sonnet 4.6, Claude API, Claude Code, and Claude.ai saw elevated error rates during a brief incident on Claude Sonnet 4.6 from 3:30 PM UTC to 3:44 PM UTC. A separate Claude.ai-related hiccup occurred around 3:27 PM UTC and lasted about a minute, contributing to intermittent failures when calling Claude APIs, Code, or Claude.ai.

Status updates show the issue was categorized as a degraded service and then resolved. The affected components returned to normal by 4:28 PM UTC after being marked degraded at 4:09 PM UTC. User reports indicated a notable spike in activity around the incident window, with recovery occurring through the late afternoon.

Users across Claude services reported slow responses and timeouts, with some requests failing to load entirely. Early activity began around 3:05 PM UTC as browsers failed to load and a surge in reports followed around 3:30 PM UTC through 3:45 PM UTC. The issue appeared to impact multiple Claude components, including Claude API (api.anthropic.com), claude.ai, platform.claude.com, Claude for Government, and Claude Code.

Users worldwide faced login issues with Claude.ai and Claude Code beginning around 1:56 PM UTC. Many reports described login timeouts, authorization failures, and disrupted Claude Code sessions, affecting the ability to sign in, start tasks, or access chats. The Claude API itself was not affected, but the user experience was significantly degraded across Claude.ai and Claude Code components.

Claude's chat service experienced a widespread outage, leaving users around the world unable to get responses or encountering errors when chatting. StatusGator detected the issue at Mar 10, 2026 5:04 PM UTC, ahead of any official acknowledgement, as reports began to surge around 5:30 PM UTC and continued through 6:15 PM UTC. Users described slow replies, timeouts, and prompts that would not load, with some noting maintenance notices as the incident progressed.

No official Claude status updates were posted during the event. The issue appeared to have global reach, with reports spanning North America, Europe, and Asia, indicating a broad impact on typical chat tasks. The outage was observed to end by 6:53 PM UTC as activity subsided and users resumed normal usage.

The Claude Desktop app became unresponsive, preventing startup and normal use for users around the world. The incident began at Mar 8, 2026 5:58 PM UTC and persisted for more than a day as engineers investigated. Users reported issues ranging from login problems to stalled workflows, with many describing difficulty in completing tasks in Claude Cowork or Claude Code.

Root cause: a daylight saving time transition created an infinite loop when the app tried to locate tasks scheduled during the skipped hour in certain time zones. The team released a fix as an update (version 1.1.5749) and provided a workaround to switch to a non-DST timezone if updating immediately. By gradually deploying the fix, Claude aimed to restore normal functionality for affected users and prevent recurrence across regions.

The service was restored by Mar 10, 2026 4:53 PM UTC, and status updates indicated the desktop client returning to normal operation. A large volume of user reports across multiple regions reflected the broad impact, underscoring how widespread the outage was for desktop users.

Users of Claude Desktop and the Claude API experienced widespread issues starting 5:18 PM UTC, with desktop apps hanging, blank screens, and slow or failed responses across Claude API (api.anthropic.com), claude.ai, and Claude Code. A notable volume of user reports across regions indicated a broad impact on core workflows that rely on Claude.

Official status updates identified the cause as a missing Filesystem connector on Claude Desktop, with a fix implemented and guidance to re-add the connector to organization allowlists to reactivate it. On 1:58 PM UTC, Claude Haiku 4.5 also showed elevated errors, and a follow-up at 4:58 PM UTC noted the Filesystem connector issue persisted but that a fix had been applied. The issue was gradually resolved, and the system returned to normal by 3:53 PM UTC.

The Claude API (api.anthropic.com) experienced a partial outage as some API requests timed out due to elevated TCP three-way handshake failures caused by network degradation at an upstream peering point. Requests that could connect were unaffected, but overall the experience degraded for developers relying on Claude's API. The incident began at Mar 6, 2026 3:38 PM UTC and was categorized as a degraded service during the event. A fix was implemented at Mar 6, 2026 4:04 PM UTC and monitoring continued; connectivity returned to normal by Mar 6, 2026 4:23 PM UTC, with the outage lasting roughly 45 minutes.

The Claude platform experienced elevated errors starting around 2:18 PM UTC, affecting Opus 4.6 and Haiku 4.5 workloads. Users reported authentication failures, non-loading models, and various errors when trying to use Claude features. The issues hit core components Claude API (api.anthropic.com), claude.ai, and Claude Code, resulting in degraded performance and intermittent access for those trying to interact with Opus 4.6 and Haiku 4.5.

Recovery unfolded in stages: a brief stabilization appeared around 2:58 PM UTC as Claude API components moved back toward up, but Haiku 4.5 problems re-emerged around 3:58 PM UTC. A fix was implemented by 4:18 PM UTC and monitoring continued. By 5:03 PM UTC all Claude components were reported up again, marking the end of the incident after roughly 2:18 PM UTC to 5:03 PM UTC (about 2 hours and 45 minutes) of degraded service. The footprint appeared global, with reports from multiple regions.

An outage disrupted Claude's Admin API and the Console, with usage reporting and cost reporting affected. Users attempting to manage accounts, access reporting screens, or run queries against the Sonnet 4.6 model reported errors and slow responses. StatusGator logged an Early Warning Signal starting at Mar 3, 2026 3:44 PM UTC, roughly 1 hour and 9 minutes before the first official acknowledgement at 4:53 PM UTC.

Engineering teams pushed fixes over the next hours. The admin API was operational again by 12:58 AM UTC and the Console (platform.claude.com) followed by 1:34 AM UTC. A separate update addressed usage reporting, and by 8:33 AM UTC the service had fully recovered. The outage affected Claude API (api.anthropic.com) and platform.claude.com, with cost reporting also impacted during the incident.