Anthropic reports engineers now ship eight times more code per quarter than 2021‑2025, with over 80% authored by its Claude model.
AI task capability length is doubling roughly every four months, progressing from 4‑minute tasks in March 2024 to 12‑hour tasks by March 2026.
Claude’s success rate on open‑ended tasks reached 76% in May 2026, a 50‑point increase in six months; Claude‑powered agents closed 97% of a safety‑research performance gap using about $18,000 compute over 800 hours.
Anthropic states it would slow or pause AI development if other frontier developers do so verifiably, and its Anthropic Institute will research verification mechanisms.
