The Claude service encountered serious technical difficulties affecting the performance of key models in the Opus and Sonnet families, as well as specialized developer tools. The incident caused disruptions to both the web interface and programmatic APIs.
What Happened
Elevated error rates were recorded in the operation of models Opus 4.8, 4.7, 4.6, Sonnet 4.6, and Haiku 4.5. The outage affected the claude.ai user interface, the API (api.anthropic.com), as well as Claude Code and Claude Cowork tools. At this time, the incident has been fully resolved, and the functionality of all models has been completely restored.
Context
The problem affected the entire Anthropic service stack, from the high-level web interface to low-level APIs and developer tools, leading to a temporary inability to use both manual and automated interactions with the models.
Why It Matters for the Industry
This incident highlights the critical vulnerability of the modern AI industry and the risk of high dependency on a single provider's infrastructure. Mass outages of flagship models expose the need to implement failover strategies and utilize multi-model redundancy to ensure business continuity.
Why It Matters for Users
For users relying on Claude via API or CLI tools, similar outages can lead to interruptions in automated workflows and service degradation. It is recommended to monitor availability status on the official Anthropic status page when performing mission-critical tasks.
What Is Not Yet Known / Limitations
There is a difference in risk assessments: ranging from purely technical availability issues to large-scale business risks for companies building their products exclusively on a single API provider.
Sources
Author
Look at AI, Editorial Team