Anthropic Experiences Recurring Outage with Claude Opus 4.6 Model
Key Takeaways
- ▸Claude Opus 4.6 experienced elevated error rates affecting claude.ai, the Claude API, platform.claude.com, and Claude Code
- ▸The issue proved to be recurring, with an initial fix implemented at 10:27 UTC failing to permanently resolve the problem
- ▸Anthropic provided regular status updates throughout the incident, which began at 06:59 UTC on March 3, 2026
Summary
Anthropic reported elevated error rates affecting its Claude Opus 4.6 model starting March 3, 2026. The incident impacted multiple services including claude.ai, the Claude API, and Claude Code. According to the company's status page, the issue was first detected at 06:59 UTC and proved to be recurring, with engineers implementing multiple fixes throughout the day.
The outage followed a pattern of investigation, identification, and attempted resolution that spanned over six hours. After initially implementing a fix at 10:27 UTC and monitoring results, the issue resurfaced by noon UTC, requiring additional intervention. As of 12:59 UTC, Anthropic confirmed they were continuing to work on a permanent solution to the problem.
This incident affected all major Claude service endpoints, potentially disrupting access for both individual users and enterprise customers relying on the API for production applications. The company maintained transparency through its status page, providing regular updates as the situation evolved.
- As of 12:59 UTC, engineers were still working on a complete fix for the ongoing service disruption
