Summary
Anthropic released Claude Code v2.1.150 with no stated user-facing features, but the same release window surfaced a widely discussed regression report that Sonnet 4.6 sessions were showing a 200K context cap instead of the expected 1M. The combination turns an otherwise quiet infrastructure release into a trust and reliability signal for heavy users.
What changed
Claude Code v2.1.150 shipped internal infrastructure changes, while community reports in the project’s public repo tied the release window to a context-window downgrade affecting Sonnet 4.6 sessions.
Why it matters
Context availability is one of the most important product promises in coding agents. Even when the root cause is still being clarified, a visible drop from marketed long-context behavior to a lower cap changes viability for large-repo and long-running agent workflows.
Evidence excerpt
The release notes say v2.1.150 contains internal infrastructure improvements with no user-facing changes, while a public repo report from the same window describes Sonnet 4.6 showing a 200K instead of 1M context limit.