On June 22, 2026, Anthropic’s Claude AI platform experienced a significant service disruption, affecting multiple flagship models and leaving developers worldwide without access for nearly 90 minutes before engineers restored full functionality.
The incident began at 00:37 UTC when Anthropic’s engineering team initiated an investigation into elevated error rates impacting several Claude models simultaneously. The disruption affected five models: Opus 4.8, Opus 4.7, Opus 4.6, Sonnet 4.6, and Haiku 4.5, marking one of the broadest multi-model outages the platform has encountered this month.
Services impacted included the entire Claude ecosystem, such as the Claude.ai chat interfaces, the Claude API (api.anthropic.com), Claude Code, and Claude Cowork. Developers relying on API integrations reported failed requests, while Claude Code users, a growing segment of enterprise engineering teams, experienced cascading workflow interruptions.
By 01:11 UTC, Anthropic identified the root cause and began implementing a targeted fix. Recovery proceeded model by model in a staged rollout:
- 01:16 UTC — Opus 4.8 recovered
- 01:33 UTC — Haiku 4.5 recovered
- 01:56 UTC — Opus 4.7 recovered
- 02:01 UTC — Fix implemented; monitoring phase initiated
- 02:06 UTC — Full resolution declared
The outage quickly drew developer frustration across social platforms, with Downdetector logging visible spikes in user-submitted reports focused on chat failures and Claude Code disruptions. For teams with production pipelines and deadline-sensitive workflows, even a sub-two-hour window of elevated errors can trigger significant downstream impact, particularly as Claude integrations deepen across software development and enterprise automation stacks.
This incident is notably the third service disruption affecting Claude this month. Prior outages were recorded on June 2, a major disruption lasting several hours that impacted Opus 4.6, the Claude API, and Claude Code, and June 5, when networking performance degradation within Anthropic’s cloud infrastructure triggered elevated error rates across models.
Despite the recurring incidents, Anthropic’s 90-day uptime figure remains above 98%, suggesting these are isolated spikes rather than systemic infrastructure failures. However, the frequency of disruptions in June alone raises questions about capacity planning as demand for Claude’s most powerful models continues to accelerate.
As of 02:06 UTC on June 22, Anthropic’s official status page confirms all components, including the Claude API, Claude Code, Claude Cowork, and Claude for Government, are fully operational. Anthropic stated the team remains in active monitoring mode to ensure stability holds following the staged recovery.
As AI platforms like Claude become increasingly integral to enterprise operations, ensuring consistent uptime is paramount. While Anthropic’s swift response and high overall uptime are commendable, the recent spate of outages underscores the need for robust infrastructure scaling to meet growing demand. Organizations relying on AI services should consider implementing contingency plans to mitigate potential disruptions in their workflows.
