On Tuesday, Anthropic experienced a significant service disruption affecting multiple Claude models, as reported on their official status page. The company announced that by mid-afternoon UTC, they had implemented a fix and were closely observing the systems for any further complications.
Incident Overview and Response
The issue was initially logged on status.claude.com as an ‘Elevated error rate across multiple models’ and was first acknowledged at 14:19 UTC. Anthropic quickly identified the root cause and began implementing a solution by 14:25 UTC. By 14:53 UTC, the fix was in place, and the company transitioned to a monitoring phase. An update at 15:28 UTC confirmed continued vigilance, suggesting core services were mostly restored, though engineers remained cautious about declaring the incident fully resolved.
Impact on Services and Products
The disruption impacted several Anthropic products, which experienced partial outages. Affected services included the claude.ai consumer interface, the Claude Console at platform.claude.com, the Claude API, Claude Code, and Claude Cowork. Notably, Claude for Government remained unaffected throughout the incident.
While the error rates were elevated, they did not constitute a total service blackout. This meant that although some user requests failed or returned errors, many were still processed successfully, preventing a complete shutdown of operations.
Recent Trends and Historical Context
This disruption is part of a broader pattern observed in Anthropic’s reliability challenges this month. The company’s status logs show a busy period for their reliability team, with repeated incidents especially affecting the Claude Opus 4.8 model.
Earlier on the same day, another incident involving Claude Opus 4.8 occurred between 06:28 UTC and 08:45 UTC. On June 22, multiple incidents were reported, impacting models such as Opus 4.8, Opus 4.7, Opus 4.6, Sonnet 4.6, and Haiku 4.5, which were resolved in the early morning UTC.
A more quantifiable event took place on June 16, when Anthropic reported a two-phase incident, with error rates for Sonnet and Opus models peaking at approximately 10 percent before stabilizing.
Stay updated on the latest developments by following us on Google News, LinkedIn, and X.
