تقييم METR داخل Anthropic: تقدّم القدرات على المهام المحددة صار تقريبًا بضعف الوتيرة السابقة، بتقدير 1.9 ضعف على مؤشر AECI، مع حذف كثير من الأدلةMETR's assessment inside Anthropic: capability progress on well-specified tasks is likely about twice as fast, with AECI growth estimated at 1.9x, and much of the evidence redacted

The independent evaluator METR published on 5 November a high-access assessment inside Anthropic, carried out from 12 August to 29 October 2026, to estimate how much AI is accelerating AI research and development itself. Its central judgment is that as of October 2026 Anthropic is likely improving its models' capabilities on well-specified tasks at around twice the rate it did between March 2024 and October 2025, with high uncertainty; METR says it would not be surprised if the true figure were closer to 1.3x or 3x.

On AECI, Anthropic's version of the Epoch Capabilities Index, benchmark scores have grown since October 2025 at an estimated 1.9 times the earlier rate, with a 95% confidence interval of 1.5x to 2.8x, implying roughly two years of historical progress in one year. METR says it is confident the acceleration was not caused by faster growth in training compute. It guesses that a majority of the current acceleration, perhaps about 80%, comes from using AI in R&D, with substantial uncertainty, and that non-AI factors likely played a large role in late 2025.

Anthropic had the right to redact any non-public information from the public report, and it redacted most of the evidence METR found most informative for its risk conclusions. Anthropic described the redactions as aimed at removing confidential information that could accelerate other AI developers. METR says the report does not achieve transparency about evidence, so the public has to rely on METR's judgment more than would be ideal.

METR did not see evidence of clear bottlenecks in algorithmic progress, training compute, experiment compute or data that would stop further progress over the next 12 months, but it says its understanding of bottlenecks changed more than once, so it is not confident in that assessment. It was told that acceleration comes mainly from AI assistance rather than fully autonomous contributions, with some signs that autonomous contributions may grow soon. METR took no payment from Anthropic and did not request access to sample non-public models.