Anthropic has confirmed that a major outage affecting its Claude AI suite—including Claude.ai, the API, Claude Code, and Claude Cowork—has been resolved following a service disruption on June 22, 2026. The incident, which began at approximately 10:19 a.m. ET, triggered widespread reports of elevated error rates and complete service failures across multiple models.
A Pattern of Instability
While Anthropic’s status page indicated a rapid response, with the company identifying the root cause and implementing a fix within minutes, the outage highlights growing concerns regarding the stability of the platform. By 10:53 a.m. ET, the company reported that it was monitoring the results of the implemented fix, noting that services were returning to normal. However, the outage did not affect the specialized ‘Claude for Government’ platform, suggesting a tiered infrastructure resilience that left standard enterprise and consumer users vulnerable.
The Shadow of Model Suspensions
The June 22 incident is compounded by the ongoing suspension of Anthropic’s flagship models, Mythos 5 and Fable 5. Since June 13, access to these specific models has been removed across all interfaces. Unlike the temporary outage, the suspension of these models remains unexplained in detail, with Anthropic citing only that the models are under “monitoring.”
Analysis: The Maturity Challenge
The convergence of these events suggests a critical juncture for Anthropic. While the company has built its reputation on safety and interpretability, the frequency of status-page updates regarding “elevated error rates” is beginning to resemble the operational struggles of other major AI providers. The decision to pull high-performance models like Mythos 5 and Fable 5 entirely—rather than simply limiting their capacity—indicates a significant underlying technical issue that exceeds typical load-balancing problems. For enterprise users and developers relying on the Claude API, the lack of transparency regarding the Mythos and Fable suspensions creates a climate of uncertainty that one-day service restorations may not fully rectify.