Anthropic Reports Elevated Error Rates on Claude Opus 4.6 and 4.7

Anthropic has posted a status update indicating elevated error rates affecting their Claude API and Claude Code services, specifically with Opus versions 4.6 and 4.7. The company is currently investigating the issue, which was first reported on May 15, 2026 at approximately 00:18 UTC.

The Claude API, accessible through api.anthropic.com, serves as the backbone for numerous applications and services that rely on Anthropic's AI models. This disruption could impact developers and businesses that have integrated Claude's capabilities into their workflows, particularly those using the Opus model which is Anthropic's flagship offering.

Claude Code, Anthropic's development tooling that integrates AI assistance directly into coding workflows, is also experiencing these elevated error rates. This suggests the issue may be affecting multiple layers of Anthropic's infrastructure rather than being isolated to a specific component.

The timing of this incident comes as Anthropic continues to compete in the rapidly evolving AI landscape. Companies like OpenAI, Google, and others are constantly iterating on their models and services, making reliability and performance critical factors for maintaining market position.

For developers and organizations relying on Anthropic's services, this incident highlights the importance of monitoring service health and having contingency plans in place when critical AI infrastructure experiences disruptions. The investigation is ongoing, with Anthropic providing updates through their status page as more information becomes available.

This situation underscores the challenges faced by AI companies as they scale their services to meet growing demand while maintaining the high reliability expectations of enterprise customers and developers who depend on these systems for critical applications.