“AI-evaluation nonprofit METR published a new report based on 11 weeks of inside access to Anthropic's internal data and staff interviews, concluding that Anthropic's AI progress is likely running about 2x as fast as before, mostly due to its own AI.”

METR's report exists, is dated 2026-11-05, and says substantially what the post says it says. The report's own summary states that benchmark scores of Anthropic's models, as measured by AECI (Anthropic's version of the Epoch Capabilities Index), have accelerated 1.9x [95% confidence interval: 1.5–2.8x] since October 2025, implying that Anthropic has made roughly two years of historical progress over the last year. On the headline framing, METR's wording is narrower than "AI progress." METR says Anthropic "is likely improving the capabilities of their models on well-specified tasks at around twice the rate" it managed from March 2024 to October 2025. On the uncertainty around that figure, METR writes that the estimate is given with high uncertainty and it would not be surprised if the true acceleration was closer to 1.3x or 3x the historical rate. On attribution, METR writes: "We guess that a majority (perhaps ≈80%) of the current (October 2026) acceleration is due to AI." This judgment is given with substantial uncertainty, but is based on a variety of lines of evidence on contributions from AI and other causes. METR also states that the acceleration in late 2025 was likely influenced by non-AI factors, and that details of many of these factors have been redacted. Separately, METR says it is confident the acceleration is not driven by an acceleration in the scale of training compute. On the access period and method, the assessment ran from August 12 to October 29, 2026, which is about 11 weeks, and was performed by METR staff at 2.5 full-time equivalents piloting a high-access evaluation at Anthropic. Press accounts describe the inputs as Anthropic's internal benchmark data, compute metrics, uplift surveys, code and inference-spend data, and about 15 staff interviews. On redactions, METR has said "We submitted a ~100-page full report on our assessment to Anthropic. For the public version, Anthropic redacted a majority of the evidence we found most informative", and that most of the redacted information was very high-level, while adding that it was not in a position to say whether the redactions were commercially justified. On the McCandlish quote, Anthropic's co-founder acknowledged the extent of the redactions in a post on X responding to METR's findings, saying "we determined that sharing exactly what is driving the speedup isn't the right thing to do, both because it's commercially sensitive and because it risks handing out a roadmap for others to go faster". On the internal models, METR states that Anthropic's internal benchmarks are increasingly saturated: Model 4 achieves greater than 50% on all the AECI benchmarks on which it was tested, and its main chart plots AECI scores of Anthropic model snapshots by the date each was available to most Anthropic employees, from Claude 3 Opus in early 2024 to internal Model 4 in October 2026. On the S-1, reporting describes new risk-factor language in the amended Form S-1/A filed 2026-11-02: "our internal estimates and our expectations regarding future models, rather than our current models, indicate that we may imminently meet our Responsible Scaling Policy's definition" of "dramatically accelerated progress attributable to AI", and that this sentence is not in the first S-1, filed October 26. The filing also states, per reporting, "We have not designated any of our current models as exceeding this threshold." On the Greenblatt quote, he wrote: "This report on AI R&D automation (RSI) is a big update towards ASI within 1-2 years. It says AI progress is sped up by RSI right now by around 2x. In AI 2027, this happens around March 2027." He added "I was seriously wrong about RSI speed up; I previously expected much less in 2026 (around 1.15x-1.3x)". The post describes him as "METR's own," which checks out: he announced "I'm joining METR to work on more investigations like our Hugging Face report" and is now listed as a Member of Technical Staff at METR, previously Chief Scientist at Redwood Research.