---
title: "METR says Anthropic's models now improve 1.9 times as fast, and Anthropic redacted the evidence"
description: "METR's eleven-week look inside Anthropic found two years of model progress packed into one, mostly from AI doing the research itself. The strongest evidence is redacted, days before an IPO investors now price nearer $1 trillion."
publisher: "The Inference"
section: "Model"
published: 2026-11-06T14:11:29.562Z
modified: 2026-11-06T14:11:29.562Z
canonical: https://theinference.org/article/metr-says-anthropic-s-models-now-improve-1-9-times-as-fast-and-anthropic-redacted-the-evidence
language: en
keywords: "Anthropic, METR, AI Safety, IPO, Model Benchmarks, AI Research"
---

# METR says Anthropic's models now improve 1.9 times as fast, and Anthropic redacted the evidence

> METR's eleven-week look inside Anthropic found two years of model progress packed into one, mostly from AI doing the research itself. The strongest evidence is redacted, days before an IPO investors now price nearer $1 trillion.

## Three models nobody outside has seen

Between July and October, [Anthropic](https://theinference.org/markets/companies/anthropic) finished training three frontier models no outsider has used, one in July, one by September, one in early October. The inventory comes from METR, a Berkeley nonprofit evaluator whose staff [worked inside Anthropic's data from August 12 to October 29](https://metr.org/blog/2026-11-05-anthropic-ai-rd-acceleration-assessment/) [1][2]. In internal Slack polls, staff put their productivity gain from the most capable hidden model near 12x, a figure METR itself calls very loose evidence [1]. That stockpile is the engine a trillion-dollar listing would sell.

## Twice the pace, and not from compute

Since October 2025, scores on AECI, Anthropic's internal version of the Epoch Capabilities Index, have climbed 1.9 times as fast, with a 95% confidence interval of 1.5–2.8x, moving from roughly 15 points a year to 29: two years of progress in one [1][3].

METR confidently rules out faster training-compute scaling, guesses about 80% of the current acceleration comes from AI doing the R&D itself, and sees no bottleneck in algorithms, compute or data for twelve months [1]. Weeks earlier, CEO [Dario Amodei](https://theinference.org/markets/people/dario-amodei) had [urged the industry to "slow the pace at which we improve the capabilities of AI models"](https://www.usnews.com/news/best-states/california/articles/2026-11-05/anthropic-is-rapidly-accelerating-ai-development-amid-slowdown-calls-outside-research-group-says) [4].

*Chart: **Six estimates of Anthropic's speedup; only the top one clears its own 2x line** Bar chart of six estimates of how many times faster Anthropic's benchmark progress became after October 2025, spanning 1.1x to 2.9x, against a reference line at Anthropic's own 2x acceleration threshold*

*Multiple of Anthropic's pre-October-2025 rate of benchmark-score growth, by estimating method. System-card figures are Anthropic's own; METR variants change the task-difficulty assumption or include rivals' models. Sources: METR, November 5, 2026; Tech Times. [1][3]*

## The proof sits behind the redactions

Anthropic held unilateral redaction rights and cut "a majority of the evidence we found most informative" for METR's risk conclusions, describing the cuts as aimed at information that could accelerate other developers [1].

METR adds that Anthropic's public August risk report understates the acceleration; [a companion review](https://www.msn.com/en-us/technology/artificial-intelligence/anthropic-ai-acceleration-hits-1-9x-on-benchmarks-metr-calls-anthropic-threshold-defense-overstated/ar-AA2gSetq) calls the company's case that it stayed under its own 2x threshold "overstated"; and the November 2 amended IPO filing says future models "may imminently meet" that threshold [3][4].

Co-founder Sam McCandlish, Anthropic's responsible scaling officer, concedes the cuts make METR's work harder to check and calls the driver of the speedup commercially sensitive [4]. Buyers are underwriting a speedometer they cannot open.

## Who pays this month

Anthropic [filed October 26 for a Nasdaq listing under ticker ANT](https://www.msn.com/en-us/news/other/michael-burry-called-the-ai-boom-catastrophically-vulnerable-following-anthropic-s-filing-ant-stock-s-valuation-is-a-more-immediate-risk/ar-AA2gPzyr) [5], with a roadshow planned for the week of November 9 and an offering some expect near $100 billion, down from early talk of $2 trillion [6].

After Thursday's $2.5 trillion tech rout, three major investors told The Wall Street Journal that $1.5 trillion is unlikely, one putting the number nearer $1 trillion [7]. Capital Group, at $10 billion, and Fidelity, at $5 billion, lead roughly $16 billion of fund money already in [8], while [Michael Burry](https://theinference.org/markets/people/michael-burry) [counts $537 billion of compute commitments](https://theinference.org/article/anthropic-s-leaked-filing-warns-of-extinction-then-signs-518-billion-of-compute-it-cannot-cancel) and calls the boom "catastrophically vulnerable" [5].

*Chart: **The IPO price tag has halved in three weeks of talk** Column chart of three reported valuations for Anthropic's IPO, falling from $2 trillion to about $1 trillion*

*Reported valuation expectations for Anthropic's Nasdaq listing, in trillions of dollars. The early figure predates last week's selloff; the latter two come from major investors speaking to The Wall Street Journal after Thursday's $2.5 trillion tech rout. Sources: Reuters via New York Post; PYMNTS. [6][7]*

## The reading that decides it

The counterweight is METR's own spread: include rival labs' models and the acceleration reads just 1.4x [3]. Anthropic must decide within days whether to launch, with October revenue still blank in the amended filing [6][7]. Next on the record are the six METR staff embedded under [the October 15 arrangement](https://www.cnbc.com/2026/10/15/anthropic-metr-securebio-ai-safety-evaluators.html), with employee-like access and publication Anthropic cannot veto outright [3][9]. Watch whether a $1 trillion price clears on conclusions alone [3][9].

## Takeaway

METR's inside look at Anthropic found model progress running 1.9 times the old rate, mostly from AI doing the research itself, but Anthropic redacted most of the supporting evidence, and investors are now pricing the IPO nearer $1 trillion than $2 trillion on evidence they cannot open.

## Sources

1. [METR, Independent Assessment of AI R&D Acceleration at Anthropic in October 2026, November 5, 2026](https://metr.org/blog/2026-11-05-anthropic-ai-rd-acceleration-assessment/)
2. [Business Insider, 3 takeaways from the big AI report on Anthropic everyone is talking about, November 5, 2026](https://www.businessinsider.com/anthropic-metr-ai-researchers-report-productivity-recursive-self-improvement-2026-11)
3. [Tech Times, Anthropic AI acceleration hits 1.9x on benchmarks; METR calls Anthropic threshold defense overstated, November 5, 2026](https://www.msn.com/en-us/technology/artificial-intelligence/anthropic-ai-acceleration-hits-1-9x-on-benchmarks-metr-calls-anthropic-threshold-defense-overstated/ar-AA2gSetq)
4. [Associated Press via U.S. News, Anthropic Is Rapidly Accelerating AI Development Amid Slowdown Calls, Outside Research Group Says, November 5, 2026](https://www.usnews.com/news/best-states/california/articles/2026-11-05/anthropic-is-rapidly-accelerating-ai-development-amid-slowdown-calls-outside-research-group-says)
5. [Barchart, Michael Burry called the AI boom catastrophically vulnerable following Anthropic's filing, November 5, 2026](https://www.msn.com/en-us/news/other/michael-burry-called-the-ai-boom-catastrophically-vulnerable-following-anthropic-s-filing-ant-stock-s-valuation-is-a-more-immediate-risk/ar-AA2gPzyr)
6. [Reuters via New York Post, Anthropic's IPO timing threatened by AI backlash, historic tech rout, November 5, 2026](https://www.msn.com/en-us/news/other/anthropic-s-ipo-timing-threatened-by-ai-backlash-historic-tech-rout/ar-AA2gQYL9)
7. [PYMNTS, Anthropic IPO Faces $500 Billion Valuation Hit After Tech Stocks Plunge, November 5, 2026](https://www.pymnts.com/news/artificial-intelligence/2026/anthropic-ipo-faces-500-billion-valuation-hit-after-tech-stocks-plunge/)
8. [Morningstar, Which Funds Have the Largest Stakes in Anthropic and OpenAI Ahead of Their IPOs?, October 22, 2026](https://www.morningstar.com/funds/which-funds-have-largest-stakes-anthropic-openai-ahead-their-ipos)
9. [CNBC, Anthropic taps nonprofits METR and SecureBio to help assess risks and safety practices, October 15, 2026](https://www.cnbc.com/2026/10/15/anthropic-metr-securebio-ai-safety-evaluators.html)


---

METR says Anthropic's models now improve 1.9 times as fast, and Anthropic redacted the evidence — The Inference. Canonical: https://theinference.org/article/metr-says-anthropic-s-models-now-improve-1-9-times-as-fast-and-anthropic-redacted-the-evidence