Ratio of Anthropic's AI capability progress indicated by leading indicators since early-to-mid 2025 to its pre-acceleration rate
- Measures
- The growth rate of Anthropic's internal leading indicators of AI capability progress (training-pipeline and related metrics, as used in its August 2026 Risk Report) over the period from the acceleration onset in early-to-mid 2025 through mid-2026, divided by the growth rate of the same indicators before that onset.
- Population
- Anthropic, including unreleased internal models
- Conditions
- Measured up to the Risk Report's coverage date (15 July 2026); total observed acceleration, not only the part attributable to AI assistance.
- Unit
- ×
Most likely about 1.90 ×; 90% interval 1.20 × to 3.00 ×.
Stated as a distribution, the precision the evidence supports. right-skewed, roughly lognormal
Assessment
The graph's estimate, all things considered: the figures and arguments it rests on, and how they were combined.
The ratio is most likely around 2, with a 90% interval of roughly 1.2 to 3.0. Anthropic's public August 2026 Risk Report described acceleration in its leading indicators starting in early-to-mid 2025 as "less than a factor of 2". METR, which reviewed the private unredacted report, calculated that the private report's own figures imply about 1.97, and its own analysis of the same indicators gave reasonable estimates between 1.3 and 2.8 depending on definition and time period. Secondary summaries round this to "about 2x". The underlying data are not public: both the indicators and METR's methodological discussion were redacted, so the estimate rests on testimony from Anthropic, which has an interest in the figure staying below its 2x early-warning threshold, and from METR, which had high access but whose analysis is also unpublished. The interval is wide because METR reports substantial uncertainty about how to measure the indicators, and because the ratio is total observed acceleration, part of which Anthropic attributes to non-AI factors and part of which may reflect catching up to the frontier. How these indicators relate to downstream capabilities is itself uncertain.
The work: which figures, how they were combined, and the arithmetic
Inputs: (1) Anthropic private risk report, implying 1.97 by METR's calculation (current-rate estimate divided by historical rate); public wording "less than a factor of 2". (2) METR's independent analysis: range 1.3 to 2.8 of reasonable estimates, presented as a span over methods rather than a confidence interval. (3) FRI and AI Futures summaries stating ~2x, derived from METR, not independent. Combination: central at 1.9, slightly below 1.97 to reflect METR's range being centred near 1.9 on a log scale (geometric mean of 1.3 and 2.8 is about 1.91). 90% interval widened slightly beyond METR's span (1.2 to 3.0) since that span is of method choices and does not include sampling error or definitional disputes. Evidence not shown: the indicator data were redacted by Anthropic.
Estimates
Each figure a source states for this quantity, in its own words, under the argument it belongs to and read against the graph's estimate.
Grey band: the graph's 90% interval, with its central value as a line. Dark marks: each source's own figure. A thin line is an interval, a thick bar its middle half, a dot a single figure, a dashed arrow a bound. An open box is where an argument places the value.
Decomposition
This claim is atomic: it bottoms out in a bedrock fact, an open empirical question, or a value premise, and does not decompose further.
Provenance
Where this quantity's evidence comes from. The figures sources state are under Estimates above.
None of its 3 sources has been traced to an origin, or to one another.
METR’s own analysis of these leading indicators, which was redacted from this report, finds a range of reasonable estimates, between 1.3 and 2.8.
Acceleration factor in Anthropic's leading indicators of AI progress; the private Anthropic report's figures imply 1.97, while the public report says 'less than a factor of 2'.
METR now estimates that capability progress at Anthropic is about 2× its earlier rate, and that around 80% of this acceleration is due to AI.
METR report of 2026-11-05; applies to Anthropic only, includes unreleased models, may partly reflect catching up, and much evidence was redacted.
they estimate that, over the past year, AI R&D progress has accelerated by ~2x relative to the trend before Oct 2025.
METR's assessment of Anthropic, based on piecewise-linear fits to Anthropic's internal capabilities index (slope 1.9x; METR would not be surprised by 1.3x to 3x).
Cite this claim: a formal citation with its evidence attached
Embed this claim: show it on another site with its current assessment
An embed reports the graph's judgment as it stands when the page is viewed: the status with its confidence, how disputed the claim is, and the date. It links back here.
<blockquote class="minerval-claim" data-claim-id="d4cfffee-9259-4d52-92b1-3b8979884218"><a href="https://minerval.ai/claims/d4cfffee-9259-4d52-92b1-3b8979884218">Ratio of Anthropic's AI capability progress indicated by leading indicators since early-to-mid 2025 to its pre-acceleration rate</a></blockquote> <script async src="https://minerval.ai/embed.js"></script>
<iframe src="https://minerval.ai/embed/claims/d4cfffee-9259-4d52-92b1-3b8979884218" title="Minerval: Ratio of Anthropic's AI capability progress indicated by leading indicators since early-t…" width="480" height="275" style="border:0;max-width:100%" loading="lazy"></iframe>
[](https://minerval.ai/claims/d4cfffee-9259-4d52-92b1-3b8979884218)
The badge now: 
Contribute
Every judgment on this page is open to challenge. A contribution is evaluated on its merits by the reviewer; if it succeeds the page changes, and if it does not, the reasons are stated. Either way the exchange becomes part of the claim’s public record.
Stated views None yet. A view stated to the graph is shown on this page in its speaker’s words, as a source like any other, and carries no weight in the assessment.
Created by extractor · Nov 13, 2026. Every judgment on this page is accompanied by a reasoning trace.