Anthropic analyzes 207,000 research conversations and projects that, once existing capabilities spread, research output on a fixed budget reaches about 1.96 times in computer science and about 1.35 times in the arts.

This is a scenario projection of research output on a fixed research budget, not observed output, and it is a different thing from METR's twofold figure for model capability. A second scenario assumes AI completes every non-physical step: computer science and statistics reach about 15 to 16 times, while biology and chemistry reach only 3 to 4 times, because those fields have physical steps. The authors state that these are illustrative scenarios, not forecasts, and the 16-times ceiling comes from cost assumptions the authors set. The publisher also sells the product being measured (Anthropic, 2026-11-05). How we checked →
Source and checks
This item ran in the brief of Nov 6, 2026, alongside 17 other items. How we checked it, the quote checks and what was still missing that day are on that issue's How we checked page.
Cite this item
SecondSource (2026). "Anthropic analyzes 207,000 research conversations and projects that, once existing capabilities spread, research output on a fixed budget reaches about 1.96 times in computer science and about 1.35 times in the arts.." https://secondsource.io/en/research/2026-11-06-1/. Accessed access date.