ANTHROPIC published a public metric framework to measure the pace of frontier
lab AI development, aiming to give outsiders clearer data on model R&D progress.
The framework targets three areas: AI participation in next‑generation AI R&D,
supervisory control over AI agent behavior, and resource investment into developing
stronger models. It includes an AI R&D Automation Index that measures Claude’s
share of company R&D. As of Aug 2026, Claude had not achieved fully autonomous
R&D in any measured domain but accounted for about 26% of ANTHROPIC’s AI R&D;
over 90% of R&D work reached ‘AI collaboration’ or higher.