METR: Anthropic AI R&D is improving about twice as fast
METR says Anthropic's AI capabilities on well-specified tasks are improving about twice as fast as the March 2024 to October 2025 rate. On AECI, progress since October 2025 is estimated at 1.9 times the earlier rate, roughly two years of historical progress in one year, and Anthropic redacted most of the evidence METR found most informative.
On 5 November 2026, METR published an independent assessment of Anthropic's AI research and development. The central claim is acceleration: as of October 2026, Anthropic is improving capabilities on well-specified tasks about twice as fast as it did from March 2024 to October 2025. METR flags that figure as subjective and says it would not be surprised if the true multiple were anywhere from about 1.3x to 3x.
The three things you need to know
1. The pace. On Anthropic's AECI benchmark, METR estimates growth since October 2025 at about 1.9 times the prior rate, with a 95% confidence interval of 1.5x to 2.8x. That is roughly two years of historical progress packed into the last year. METR says this was not driven by faster training-compute scaling.
2. The cause. METR guesses that a majority, about 80%, of today's acceleration comes from AI assisting R&D. That help is mostly people working with AI, not systems that are already fully autonomous. The assessment is subjective and based on evidence METR reviewed inside Anthropic.
3. The catch. Anthropic redacted most of the evidence METR found most informative from the public write-up of the roughly 100-page report, so outsiders cannot independently check the numbers. Looking ahead, METR saw no clear bottlenecks that would stop further acceleration over the next 12 months, while stressing that this outlook is uncertain.
Why it matters
If AI systems are already speeding up their own research pipeline, the gap between what labs can demonstrate and what outsiders can verify gets wider. Acceleration without transparent evidence is hard for governments, customers and researchers to price. METR's related review of the R&D section in Anthropic's August 2026 risk report sits next to this assessment as further context on how Anthropic describes the same period.
Dany's take
The 2x figure is the headline, but the redactions are the story. METR is one of the few groups that gets this kind of access, and even then the public only gets the summary. Treat the acceleration claim as a serious signal, not a settled measurement, and watch whether other labs invite the same kind of outside review with fewer black bars.
Sources
Source: METR