Anthropic's API entered degradation mode at 10:30 AM UTC-5 today, marking the fourth incident in two weeks. The outage affects enterprises relying on Claude models for production workflows including chat interfaces, code generation, and automated billing systems.
The timing is notable. Similar short degradations hit on January 28 (5 minutes) and January 26 (10 minutes). On February 1, API errors started at 11:36 UTC and resolved by 11:55 UTC, a 19-minute window. January 22 saw a larger spike with roughly 3,000 users reporting issues on Downdetector before Anthropic rolled out a fix.
Monitoring data from Updog.ai shows 884+ outages tracked over the past year, with recent incidents ranging from 5 minutes to nearly 4 hours. The pattern suggests infrastructure strain rather than isolated failures. StatusGator reported the service operational as of February 2 at 2:38 PM PST, though user reports continued.
What this means in practice: enterprises building on Claude's API face reliability questions. Anthropic positions itself on AI safety and reliable systems, but the API's track record shows persistent availability gaps. CTOs evaluating Claude for production use should factor in these degradation windows, particularly for customer-facing applications where 25-minute outages compound.
The company's status page (status.anthropic.com) typically acknowledges incidents within 15-30 minutes, per historical data since June 2024. Recent Claude Code releases addressed authentication, OAuth, and proxy issues in January and February, suggesting ongoing infrastructure work.
Three things to watch: whether today's degradation extends beyond the typical 5-25 minute window, if Anthropic provides root cause detail (they rarely do), and whether the two-week pattern indicates deeper scaling challenges. Past incidents hit Opus models and login systems hardest.
The API remains in degradation mode as of this writing. For teams with Claude dependencies, the real test is how quickly your fallback systems kick in.