Anthropic reported elevated errors affecting its Opus 5 model across multiple services on July 26, 2026. The incident began at 09:17 UTC when the company started investigating the issue. By 09:45 UTC, the root cause was identified and a fix was being implemented. Engineers continued working on the solution through 10:28 UTC, and a fix was deployed by 10:34 UTC. Full resolution was confirmed at 10:44 UTC after monitoring confirmed stability.

The outage impacted claude.ai, Claude Console (platform.claude.com), Claude API (api.anthropic.com), Claude Code, and Claude Cowork. Users experienced degraded performance or failures during the approximately 90-minute window. Anthropic did not disclose the specific technical cause but followed its standard incident response process: detection, investigation, identification, remediation, and verification.

The incident was tracked on Anthropic's public status page, which provides real-time updates for developers and enterprise customers relying on the Claude API for production workloads. Opus 5 is the latest iteration of Anthropic's flagship language model series, and the elevated errors affected both free and paid tiers. No data loss or security breach was reported.

Service health has since returned to normal across all affected platforms.