Claude.ai, the AI platform from Anthropic, reported a surge of 503 errors across its web, mobile, and API services on April 15, 2026. The outage disrupted login, voice mode, and workspace creation, forcing users to retry repeatedly.

Investigators traced the problem to a misconfiguration in the API gateway that throttled requests to non‑Opus models. A patch merged into the deployment pipeline on April 14, 2026, and monitoring teams began verifying restoration.

Previous incidents in early April highlighted similar login failures and elevated error rates for Sonnet 4.6 and Opus 4.6. Each fix involved rolling updates and real‑time performance checks.

Anthropic’s rapid response and transparent incident timeline illustrate the challenges of scaling conversational AI services, underscoring the need for resilient architecture and real‑time observability.