🔍 Read the full analysis: What METR’s Research Says About AI And The Pace Of Development on ThorstenMeyerAI.com
Get business pricing on monitors, keyboards and dev gear
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
METR’s November 5 assessment estimates that Anthropic’s progress on specified AI capability tasks accelerated about 1.9 times after October 2025, with AI use likely responsible for much of the change. The estimate is based on internal company data, but Anthropic redacted much of the evidence METR considered most informative, leaving important limits on public verification.
METR, a nonprofit that evaluates frontier AI systems, estimated that Anthropic’s progress on AI capability tasks accelerated about 1.9 times after October 2025, in an assessment published November 5. The report says AI assistance likely accounts for much of the increase, but Anthropic redacted a majority of the evidence METR found most informative, limiting what outsiders can independently check.
METR researchers worked with Anthropic’s internal data for 11 weeks, from August 12 to October 29, 2026. The assessment drew on benchmark results for unreleased models, training metrics, staff surveys and about 15 interviews. METR said roughly 2.5 full-time-equivalent staff took part, working on Anthropic devices. The nonprofit said it received no payment; Anthropic supplied laptops and free model tokens and retained the right to redact non-public material.
METR’s main estimate compares the rate of change in Anthropic’s capability index before and after a statistical break point it places at October 7, 2025. The index rose by about 15 points per year before the break and about 29 points per year afterward, yielding an estimated 1.9x rate. METR gives a 95% confidence interval of 1.5x to 2.8x and says it would not be surprised by a result between 1.3x and 3x.
The report attributes perhaps 80% of the acceleration to AI use, while saying it is confident the change did not result from faster growth in training compute. That attribution is METR’s estimate, not a directly measured share. Anthropic’s own figures, cited in the report, say Claude led 26% of the company’s AI research and development work in August 2026, compared with under 1% in February. Anthropic also reported about 30,000 agents operating at a time on its main internal platform.
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
Why AI-Assisted Research Pace Matters
The report addresses whether AI is helping produce more capable successors, a question with implications for both the speed of AI development and the time available to evaluate its risks. If AI tools make research and engineering progress faster, labs and policymakers may have less time to assess new capabilities and put safeguards in place. METR’s estimate is evidence of a possible acceleration at one company, not proof that the whole industry is moving at the same rate.
Anthropic’s Responsible Scaling Policy sets a threshold for what it calls automated research and development: a doubling of progress relative to a baseline, where the increase is plausibly substantially attributable to research or engineering automation. METR says it is plausible that Anthropic crossed its early-warning threshold by July 15. Separately, Anthropic wrote in a November 2 SEC filing that it “may imminently meet” its definition of dramatically accelerated progress. These statements make the estimate relevant to the company’s own safety-policy triggers, though neither establishes that fully autonomous AI research has arrived.
The evidence-access issue also matters. METR’s work is unusual because evaluators examined internal data rather than testing only released models. Yet the redactions mean readers cannot inspect much of the material behind the assessment. The result offers more detail than an outside benchmark alone, but less transparency than a fully public audit.
Products Worth Considering
Apple MacBook Air 13-inch Laptop with M5 Chip, 16GB Memory, 512GB SSD, 12MP Center Stage Camera, Touch ID, Wi-Fi 7, Midnight
Samsung Galaxy Tab A7 Lite 8.7-Inch 32GB Wi-Fi Tablet, Gray
Apple iPad mini (A17 Pro), 128GB, Wi-Fi 6E, Space Gray
As an affiliate, we earn on qualifying purchases.
How METR Defines Acceleration
METR defines a 2x acceleration as making two years of historical progress in one year. Its measure combines performance across benchmarks into a capability index; it does not mean researchers are twice as productive in every part of development. Model training, testing, deployment and safety work can still take time even if performance on measured tasks improves faster.
The 1.9x figure is specific to Anthropic’s models and METR’s chosen index and time periods. When METR instead examined frontier models from all developers, it estimated a smaller change: progress rose from about 17 to 24 index points per year, or 1.4x. The report says this difference could mean Anthropic was behind the frontier in 2025, making its subsequent catch-up look unusually fast, or that its earlier models were less tuned to benchmarks.
METR’s estimate also varies with assumptions about task difficulty, ranging from 1.1x to 2.9x in one set of checks. The report says most other checks stayed near 2x. Anthropic’s own Haiku 5.5 system card reported a 1.7x estimate; METR says its higher result mainly reflects the inclusion of newer, unreleased models.
“A 2x acceleration means that 2 years of historical progress is made in 1 year.”
— METR
What the Redactions Leave Unknown
The public report does not disclose much of the evidence METR judged most informative: Anthropic redacted a majority of it. METR also dropped a planned examination of how the company divides compute between safety work and capability development. The available material therefore does not let readers fully reproduce the analysis or assess every underlying indicator.
The 1.9x estimate is not a direct measure of autonomous AI research, nor does it establish that AI systems can independently set goals or develop their own successors. METR’s estimate that AI accounts for about 80% of the acceleration remains uncertain, as does how much the Anthropic-specific result reflects catching up with other labs, benchmark tuning or other factors. The source material does not provide a detailed account of the report’s findings beyond the metrics and caveats described here.
Further Evidence and Policy Triggers
The next useful developments would be additional public data that let independent researchers test METR’s estimates, and any update from Anthropic on whether it considers its policy threshold to have been reached. The company’s November 2 filing said it may imminently meet its definition; it did not, in the material cited here, say that the threshold had been formally crossed.
Readers should also watch for further METR analysis and comparable assessments of other AI developers. Until more underlying evidence is available, the 1.9x figure is best treated as a carefully qualified estimate for Anthropic over a defined period—not a settled measure of the pace of AI development across the industry.
Key Questions
What did METR estimate?
METR estimated that Anthropic’s capability progress on its selected measures accelerated about 1.9x after a break point it places at October 7, 2025. Its 95% confidence interval is 1.5x to 2.8x.
Does the report say AI is building AI on its own?
No. The report describes AI assisting people with research and engineering. It does not establish that an AI system can autonomously set goals and design and develop its own successor.
How much of the acceleration does METR attribute to AI?
METR estimates that perhaps 80% of the acceleration is due to AI use. The report presents this as an uncertain judgment, not a directly measured proportion.
Why can’t the public fully verify the result?
Anthropic redacted a majority of the evidence METR found most informative. METR also did not complete a planned check of how Anthropic splits compute between safety and capability work.
Does the estimate apply to every AI company?
No. The headline figure concerns Anthropic. METR’s comparison across frontier developers found a smaller estimated acceleration of 1.4x, and the report notes that both figures depend on measurement choices and assumptions.
Source: ThorstenMeyerAI.com
