🔍 Read the full analysis: A Closer Look At METR’s Research On AI Development At Anthropic on ThorstenMeyerAI.com
Get the latest gadgets delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
METR’s November 5 assessment estimates that Anthropic’s progress on specified AI capability tasks accelerated 1.9 times after October 2025, with a 95% confidence interval of 1.5x to 2.8x. METR attributes much of the change to AI use in research and engineering, but says Anthropic redacted a majority of the evidence it considered most informative.
METR, a nonprofit that evaluates frontier AI systems, estimated on November 5 that Anthropic’s progress on specified AI capability tasks accelerated 1.9 times after October 2025. The outside assessment drew on 11 weeks of access to internal company data, but METR said Anthropic redacted a majority of the evidence it found most informative, limiting what readers can independently examine.
METR’s central estimate compares Anthropic’s pace of capability gains before and after a change point it places at October 7, 2025. Its measure, AECI, combines scores across benchmarks into a single capability index. METR estimates the index rose by about 15 points per year before the change point and about 29 points per year afterward. It describes that difference as roughly two years of historical progress in one year.
The estimate has a 95% confidence interval of 1.5x to 2.8x. METR says it would not be surprised by a result between 1.3x and 3x. Under different assumptions about task difficulty, its estimate ranges from 1.1x to 2.9x; the organization says most of its other checks remain near 2x. The report’s figure is higher than the 1.7x estimate in Anthropic’s Haiku 5.5 system card, partly because METR included newer models that were not publicly released.
METR’s assessment used benchmark results for unreleased models, training metrics, surveys and about 15 staff interviews. The work involved roughly 2.5 full-time-equivalent staff working on Anthropic devices from August 12 to October 29. METR says it received no payment. Anthropic supplied laptops and free model tokens and retained the right to redact non-public information.
Infographic 01 · How fast is AI speeding up AI?
About twice the old pace, most likely because of AI
An outside evaluator's estimate from internal Anthropic data, plus Anthropic's own measurements. Wide uncertainty, much of the evidence redacted.
Acceleration of Anthropic’s benchmark progress (AECI) since October 2025, 95% confidence interval 1.5x to 2.8x. METR’s overall judgment: around twice the old rate; it “would not be surprised” by 1.3x or 3x.
Capability points per year
AECI trend of Anthropic models, break point 7 Oct 2025
All labs before October 2025: +17 a year, so 1.4x. Different task-difficulty assumptions: 1.1x to 2.9x.
Sources: METR assessment, 5 Nov 2026 · METR risk report review · Anthropic Institute measurements.
Powered by Thorsten Meyer AI
What a Faster R&D Pace Could Mean
The estimate bears on whether AI systems are helping researchers build and improve later systems quickly enough to change the pace of AI development. If that contribution grows, decisions about monitoring and safeguards may need to account for AI-assisted research and engineering, not only the capabilities of released models.
METR estimates that perhaps about 80% of the acceleration comes from AI use, while saying it is confident the change did not result from faster growth in training compute. That attribution is an assessment, not a directly established causal finding. The distinction matters: a measured increase in benchmark progress does not by itself show that AI systems are autonomously designing their successors, or that all stages of development are moving at the same speed.
The report also speaks to the limits of outside scrutiny. METR had unusually deep access compared with evaluators who can test only released systems, but redactions leave readers unable to inspect much of the evidence behind its conclusions. The result is a significant independent assessment whose headline estimate is more informative than public benchmark data alone, but whose supporting record is incomplete.
Products Worth Considering

Apple Watch SE 3 GPS 40mm Smartwatch

Sonos Arc Ultra Dolby Atmos Soundbar, Black

Amazon Kindle Scribe Colorsoft 64GB with Premium Pen
As an affiliate, we earn on qualifying purchases.
Anthropic’s Twofold Progress Trigger
Anthropic’s Responsible Scaling Policy, version 3.4, took effect on July 8, 2026. It identifies two possible triggers for what it calls automated research and development: a model that could replace the company’s entire research staff at competitive cost, or progress doubling in a way plausibly attributable in substantial part to research or engineering automation. The policy defines a doubling as achieving in one year the progress otherwise expected over two years.
Anthropic’s policy describes this threshold as an early warning, not proof that a particular danger has arrived. In a November 2 SEC filing, the company wrote that it “may imminently meet” its definition of dramatically accelerated progress. METR says it is plausible that the threshold had already been crossed by July 15, but that is the evaluator’s judgment rather than a confirmed determination by Anthropic.
METR’s comparison of Anthropic alone found a larger acceleration than its comparison across frontier developers: the broader measure rose from about 17 to 24 index points per year, or 1.4x. METR says the difference could reflect Anthropic catching up from a slower position in 2025, or earlier models being less tuned to the benchmarks. These alternative explanations complicate any claim that the company’s pace alone represents a general change across the field.
“A 2x acceleration means that 2 years of historical progress is made in 1 year.”
— METR
Redactions Limit Independent Checks
The report’s most important limitation is that Anthropic redacted a majority of the evidence METR found most informative. The source material also says a planned examination of how the company divides computing resources between safety work and capability development was dropped. The public record therefore does not permit a full independent check of the underlying observations or that allocation.
The 1.9x figure is an estimate tied to METR’s index, choice of models and assumptions about benchmark task difficulty. The broader frontier comparison is lower, and METR itself gives plausible explanations for why its Anthropic-only figure could overstate or understate the change. It remains unclear how much the measured acceleration reflects improvements in real-world research output, how consistently AI tools are used across teams, and whether the pace will persist.
The account also does not establish that Anthropic has achieved fully autonomous recursive self-improvement. The assessment concerns AI helping people with development work. The scale of that help and its causal contribution remain less certain than the reported benchmark trend.
Further Evidence on Automation
The immediate next step is further disclosure or independent evaluation that could test METR’s estimate against evidence not available in the public report. Readers will also be watching for any updated statement from Anthropic about whether it considers its Responsible Scaling Policy’s progress trigger met, and what safeguards would follow if it does.
Future measurements will need to show whether the estimated rate continues, changes, or was specific to the period and benchmarks METR examined. METR’s report provides a dated assessment based on data collected through October 29, 2026; it does not establish the company’s current pace beyond that period. Until additional results or fuller supporting evidence emerge, the 1.9x estimate should be read as an independent but partly uncheckable assessment, not a settled account of AI development across the industry.
Key Questions
What did METR estimate?
METR estimated that Anthropic’s progress on its specified capability measure accelerated 1.9x after a change point it places at October 7, 2025. Its stated 95% confidence interval is 1.5x to 2.8x.
Does the report show that AI is building AI autonomously?
No. The assessment concerns AI helping people with research and engineering. Anthropic says it has not reached fully autonomous development of a successor, and METR’s estimate does not establish that it has.
Why is the estimate difficult to verify?
METR says Anthropic redacted a majority of the evidence it considered most informative. The report also dropped a planned check of how computing resources are divided between safety and capability work.
What did METR say caused the acceleration?
METR estimated that perhaps about 80% may be due to AI use and said it was confident the increase did not come from faster growth in training compute. The 80% figure is an attributed assessment, not a directly measured causal result.
Has Anthropic confirmed it crossed its policy threshold?
The supplied material does not establish that Anthropic formally declared the threshold met. A November 2 filing said the company may imminently meet its definition, while METR said crossing it by July 15 was plausible.
Source: ThorstenMeyerAI.com
