METR estimates AI tools helped accelerate Anthropic’s model development
AI · First reported 5 Nov, 18:02 · Updated 18:19
An 11-week METR assessment estimated that Anthropic’s model capabilities on well-specified tasks improved about 1.9 times faster after October 2025 than in the earlier period it examined. METR said the estimate is uncertain, and that AI was not yet making highly significant, fully autonomous contributions to the most important stages of model training.
- METR estimated a 1.9-times increase in Anthropic’s improvement rate on well-specified tasks after October 2025, while stressing substantial uncertainty.
- The assessment drew on confidential research, interviews and performance data from unreleased models; METR said faster growth in training compute did not explain the acceleration.
- METR said AI was not yet making highly significant, fully autonomous contributions to the most important stages of model training.
Summary written by Lux AI from 2 articles
Key facts
- Assessment period
- 11 weeks, between August and October
- Estimated acceleration
- 1.9 times the previous rate for benchmark capabilities on well-specified tasks
- Comparison periods
- After October 2025 versus March 2024 to October 2025
- METR’s stated uncertainty range
- The true rate could be about 1.5 to 2.5 times the historical rate
- Compute finding
- METR said acceleration was not due to faster growth in training compute
Source spectrum
Far left
Left
Lean left
Centre
Lean right
Right
Far right
Unrated
Left 0 · Centre 1 · Right 0 of 1 rated sources · +1 unrated
Images from the coverage
Background
All coverage
Coverage over time
LeftCentreRight
Articles per hour since first reported, by side.
Ownership
- Unknown 50% (1)
- Government 50% (1)
Share of the 2 sources by owner type.