METR subjectively estimates Anthropic's models are gaining capability about twice as fast as from March 2024 to October 2025, and says Anthropic redacted most of the evidence METR found most useful

Concept illustration of two side-by-side panels, each framed with a black top bar lettered METR. In the left panel, a slope rises from left to right along a baseline, a shallow beige wedge up to the middle and then a steeper brick-red section ending in a flat top on the right, where a laptop lettered ANTHROPIC sits; in the right panel, below a sign lettered ANTHROPIC, a sheet headed EVIDENCE shows thick black bars among thin grey lines, and a standing person with dark wavy hair, to the right of the sheet, holds a black marker with its tip at the right end of one of the bars. Printed on the image: About 2x past capability pace, METR subjective estimate.

METR, a nonprofit that evaluates AI capabilities, published two reports on November 5 assessing whether AI has sped up Anthropic's AI R&D (METR assessment, 2026-11-05; METR review, 2026-11-05). Our inference: METR says Anthropic redacted most of the evidence METR found most useful, so companies vetting Anthropic currently depend on METR's word on whether Anthropic's models crossed the company's own threshold for AI-driven R&D speedup.

METR's roughly twofold figure measures capability growth, which is not yet the number Anthropic's threshold asks for. METR judges that, as of October, Anthropic's models were gaining capability on well-specified tasks at about twice the rate of March 2024 to October 2025, and that 1.3 times or 3 times would not be a surprise. That is the subjective judgment of most METR staff on the project. METR guesses that about 80% of the current acceleration comes from AI. The 80% is a share of the acceleration, not of overall speed. Anthropic's own safety rules, the Responsible Scaling Policy, set a threshold with two conditions: capability progress at double the pace seen without significant AI contributions, and a doubling attributable to the automation of R&D. METR states that the assessment doesn't pay special attention to this threshold.

Anthropic assessed whether its models had crossed this threshold as of July 15. The public version of Anthropic's August risk report is more confident than the private version that the models had not crossed the threshold. The first yardstick is leading indicators, which reflect acceleration earlier than model capability does. The public version says these indicators are below a factor of two. METR calculates from the private version that one of them is about 1.97 times, and considers "around 2" the better description. The second yardstick compares Anthropic's pace of progress with the pace it would have without AI assistance. The public version says Anthropic isn't yet at twice that pace. The private version says slightly less than twice, and also that exceeding twice is plausible. METR considers a crossing plausible at the time, but most relevant researchers at METR and Anthropic still lean toward no crossing. METR also agrees that catastrophic risk is low.

These multiples can't be recomputed from public data, because Anthropic holds unilateral redaction rights over non-public information. METR writes that Anthropic redacted most of the evidence METR found most useful, on the grounds that it could help other developers. METR writes that outsiders therefore rely on METR's judgment more than would be ideal. Anthropic's November 2 amended IPO filing says that, based on internal estimates and expectations for future models, the company may imminently meet this threshold. Anthropic has not designated any current model as exceeding it. The same passage says Anthropic has in the past identified an acceleration only after it occurred (Anthropic S-1/A, 2026-11-02).

Not proven yet: both reports share one set of authors, with no second independent line of evidence How we checked →

Expert view · Daniel Kokotajlo (co-author of the AI 2027 scenario; interests: no public disclosure found): "IF (big IF) Anthropic really is imminently about to meet the RSP threshold of 2x acceleration over baseline, then it seems timelines to full RSI and superintelligence may be very short indeed." (Daniel Kokotajlo, 2026-11-03) That post predates the METR reports.

Next to watch: the next amendment to Anthropic's IPO filing and the follow-up report the filing says will be published. If by the end of March 2027 (a deadline SecondSource set) a body other than Anthropic publicly rules on whether Anthropic's models crossed this threshold, using evidence outsiders can check, the inference that companies must depend on METR's judgment changes. METR's ruling this time, based on non-public evidence, doesn't count.

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). "METR subjectively estimates Anthropic's models are gaining capability about twice as fast as from March 2024 to October 2025, and says Anthropic redacted most of the evidence METR found most useful." https://secondsource.io/en/news/2026-11-06-1/. Accessed access date.