Anthropic Outage History

Claude Haiku 4.5 experienced elevated errors and latency for 47 minutes, affecting claude.ai, console.anthropic.com, and api.anthropic.com. Anthropic implemented a fix and monitored the results before confirming the incident was resolved.

Claude.ai desktop application users experienced unavailable connectors for 2.9 hours due to affected connectors being removed from organization allowlists. The issue was resolved by allowing manual re-addition of connectors while Anthropic works to automatically restore the impacted allowlists.

Anthropic experienced elevated error rates with Opus 4.6 and Sonnet 4.6 models for 2.5 hours, affecting claude.ai, the developer console, and API services. The issue was identified and a fix was implemented. The incident was fully resolved after the engineering team completed their remediation efforts.

Claude Opus 4.6 experienced elevated error rates starting at 1:33 PT, affecting both claude.ai and api.anthropic.com services. Error rates returned to baseline levels at 2:19 PT, and the incident was fully resolved after 54 minutes of disruption.

Claude Desktop version 1.1.9310 had a major issue where Dispatch sessions stopped responding - messages were received and processed but replies did not appear in conversations, while standard Cowork sessions remained unaffected. The incident lasted 3.9 hours and was resolved by fixing the underlying issue that prevented agents from sending responses and releasing an updated desktop app version 1.1.9493.

Anthropic experienced elevated errors on Opus 4.6 Fast Mode that affected both the console and API services for 1.9 hours. The company investigated the issue, implemented a fix, and monitored the results before confirming full resolution.

Anthropic experienced elevated error rates affecting Opus 4.6 and temporarily Sonnet 4.6, impacting claude.ai, console.anthropic.com, and api.anthropic.com for 9.8 hours. Sonnet 4.6 was restored to normal operation after 5 hours, while Opus 4.6 error rates were gradually reduced from severe levels to baseline over the incident duration. The issue was fully mitigated and resolved after implementing both temporary fixes and a permanent solution.

Anthropic experienced elevated errors affecting MCP (Model Context Protocol) calls on claude.ai for 19 minutes. The issue impacted some users' ability to make successful MCP calls to the service. The incident was resolved after investigation and monitoring.

Anthropic experienced elevated errors affecting Claude Opus 4.6 and Sonnet 4.6 requests across claude.ai, console.anthropic.com, and api.anthropic.com for 2.9 hours. The issue was identified as causing a low rate of errors on requests to these specific Claude models. The incident was fully resolved after the team worked to address the underlying problem.

Anthropic experienced elevated errors on Claude Opus 4.6 that affected claude.ai, console.anthropic.com, and api.anthropic.com for 9.2 hours. The incident involved multiple rounds of investigation and mitigation attempts, with errors recurring after an initial fix was implemented. The issue was ultimately resolved after extended monitoring confirmed the final mitigation was successful.