Anthropic·anthropic.com·2025-05-22

Claude 4 release notes

71/ 100 · AI-ready

Strong atomic structure but heavy on narrative framing. Benchmarks are present but buried inside paragraphs.

View original ↗
What it gets right
  • Versioned, dated announcement at a stable URL
  • Capabilities described as discrete behaviors
What it misses
  • Benchmarks not in tabular or list form
  • No machine-readable changelog
The fix

Promote the benchmark grid to a table and publish a parallel /changelog.json feed.

Built by High Caliber AI

We do this for brands. Want us to do it for yours?

AI-native press releases, GEO audits, and answer-engine strategy for companies that want to be cited — not skimmed.