Anthropic says Claude now "leads" 26% of its frontier model research — but the metric is fuzzy
The Decoder · rss · 2026-09-18
For the first time, Anthropic is releasing metrics on how it uses AI to build AI: Claude now "leads" 26% of the work on future models, up from under 1% in February. The Decoder cautions that the underlying scale is fuzzy, the scoring comes from Claude itself, and "lead" means less than it sounds.
More from Companies & People
- Meta Built an AI Agent as a Domain 'Secondary Expert' — With Auditable Knowledge Architecture — Meta_Engineers · 2026-09-18
- Cryptographer Matthew Green: OpenAI security is 'an aspiration,' not an actual security team — matthew_d_green · 2026-09-18
- Publicis Sapient CEO: Enterprise AI adoption is like bringing electricity to the outside of an old factory — iamrobotbear · 2026-09-18
- UCSD math dept split courses into real and fake versions with identical degree credit — sytelus · 2026-09-18
- Internal Emails Call AI Training 'Largest Theft of Labor in History,' Undercutting OpenAI's Fair Use Defense — The Decoder · 2026-09-18
- 25 early-career engineering roles open now: OpenAI pays $180K+, Sierra hiring agent engineers — lennysan · 2026-09-18