Elevated error rates to Sonnet 4.5 on Claude Code (status.claude.com)

Anthropic’s Claude Code experienced elevated error rates after an update to Sonnet 4.5; the incident went through investigation, a fix was deployed, and the service is now resolved but still being monitored for regressions. The public status timeline shows the issue moved from “Investigating” to “A fix has been implemented and we are monitoring the results,” and finally “Resolved,” with ongoing monitoring to ensure stability. The outage specifically affected Claude Code, Anthropic’s code-oriented assistant, rather than general Claude chat models. For the AI/ML community this matters because code models are widely used in developer tooling, CI pipelines, and production systems where increased error rates translate directly into developer friction and potential automation failures. Elevated errors around a model release can indicate regressions in inference serving, model serialization, routing/traffic policies, or compatibility between model weights and serving infra. Practically, users should verify any recent calls that hit Claude Code (retry/backoff, idempotency, logging), watch for related release notes, and be prepared for follow-up patches. Operators and engineers should also expect post-incident root-cause disclosures to learn whether this was a model-logic regression, serving bug, or deployment/configuration issue that could inform hardening and observability best practices.