RealTeaNewsAI, Tech & CryptoNews

METR says Anthropic AI capabilities improving at roughly twice prior rate

Independent evaluators estimate about 1.9x acceleration on Anthropic internal benchmarks since October 2025, with most current acceleration attributed to AI used in R&D — while Anthropic redacted much of the evidence.

Published

RTN graphic: METR Anthropic AI R&D acceleration assessment
RTN graphic card. Not a METR chart. RealTeaNews graphic.

METR, an independent AI evaluation group, says Anthropic is likely improving model capabilities on well-specified tasks at around twice the rate seen between March 2024 and October 2025 — while warning that Anthropic redacted most of the evidence METR found most informative, according to METR’s 5 November public assessment.

Headline numbers

METR estimates Anthropic’s AECI benchmark trend has accelerated about 1.9x since October 2025 (95% confidence interval 1.5–2.8x). Staff say they would not be surprised if the true figure were closer to 1.3x or 3x. They are confident the acceleration was not driven by faster training-compute scaling.

METR’s subjective judgment is that a majority — perhaps around 80% — of the current (October 2026) acceleration is due to AI used inside Anthropic’s R&D, with substantial uncertainty and non-AI factors important in late 2025.

The redaction problem

METR states Anthropic redacted a majority of the evidence it considered most informative, describing the cuts as aimed at confidential information that could accelerate other developers. The public therefore has to rely more on METR’s judgment than METR considers ideal.

Bottlenecks

METR says it did not identify clear bottlenecks that would prevent further acceleration over the next 12 months, while noting its view of bottlenecks changed several times during the review and remains uncertain. That is a risk statement, not a prediction that runaway self-improvement has begun.

What we do not know

  • The redacted internal metrics.
  • How AECI maps to real-world impact.
  • Whether other labs show the same pattern.

Nobody was contacted. RTN reports METR’s published conclusions and caveats.

Sources

  1. Independent Assessment of AI R&D Acceleration at Anthropic in October 2026, METR. Accessed 6 November 2026.
  2. Anthropic is rapidly accelerating AI development amid slowdown calls, outside research group says, Associated Press. Accessed 6 November 2026.

Sources link to external sites. Their content is the responsibility of their publishers.

More in AI, Tech & Crypto