Anthropic's AI-attributable AI-capability progress rate in November 2026, relative to its pre-October 2025 rate
- Measures
- The ratio of Anthropic's rate of AI-capability progress in November 2026 to the rate it would have had without AI assistance to its R&D, expressed relative to its pre-October 2025 trend rate (the part of the current acceleration attributable to AI rather than to compute growth, staffing or other factors). A value of 1 means no AI-attributable acceleration.
- Population
- Anthropic's overall frontier AI R&D, as tracked by capability indices such as Anthropic's internal Epoch Capabilities Index (AECI)
- Conditions
- Instantaneous or recent-average rate as of November 2026; baseline is the pre-October 2025 trend (roughly the fastest extended progress without significant AI contributions to AI R&D).
- Unit
- ×
Most likely about 1.60 ×; 90% interval 1.15 × to 2.60 ×.
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 best estimate is that AI assistance has sped Anthropic's capability progress to about 1.6 times its pre-October 2025 rate as of November 2026, with a 90% interval of roughly 1.15x to 2.6x. The only direct evidence is METR's assessment of AI R&D acceleration at Anthropic, which put overall acceleration at about 2x (plausible range 1.3x to 3x) and guessed that about 80% of it is due to AI; the AI Futures Project reads this as about 1.7x attributable to AI. The 2x figure has partial public support: on Anthropic's internal capability index, the trend fitted after October 2025 has 1.9 times the slope of the earlier trend. The 80% attribution has none, because Anthropic redacted most of the evidence METR found most informative. That makes the AI-attributable figure largely evidence not shown, from an examiner whose access Anthropic controlled, although METR is an independent organization and its reported prior timelines were not aggressive.
Several considerations pull in opposite directions. The breakpoint was chosen for best fit, a one-time capability jump can masquerade as a slope change over a short window, and an internal benchmark index is one instrument; these push toward the lower end. Anthropic's own system card judged that its Mythos 5.2 model does not double the internal rate of aggregate progress relative to an unassisted counterfactual, which is consistent with a figure below 2x but not informative about how far below. Against that, the trend fit averages a full year, and if AI's contribution has been growing, as METR believes, the rate in November 2026 would exceed the year average. The estimate therefore sits a little below the 1.7x reading, to allow for the unverifiable attribution share and the fit's selection effects, with a long right tail for continuing acceleration.
The figure would move most with release of the redacted evidence on attribution, further data points on the internal capability index, or an independent replication on public benchmarks.
The work: which figures, how they were combined, and the arithmetic
Inputs: METR total acceleration ~2x (stated plausible 1.3x to 3x); AECI post/pre-Oct-2025 slope ratio 1.9x; METR attribution 'perhaps ~80%'. AI Futures conversion: 1 + 0.8*(2-1) = 1.8 or 2^0.8 = 1.74; they report 1.67 to 1.74x depending on reading, rounded to 1.7x. My adjustments: (a) attribution share uncertain and redacted, treat as roughly 0.5 to 0.95; (b) total acceleration taken with METR's range 1.3 to 3x, centered slightly under 2 given breakpoint selection; (c) instantaneous rate possibly above year average (right tail). Combining: at total 1.3x and 60% attribution, AI-attributable about 1.17x; at 3x and 90%, about 2.7x. Median shaded from 1.7 to 1.6. Quantiles: 5% 1.15, 25% 1.4, 50% 1.6, 75% 1.85, 95% 2.6.
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.
Its one source has been traced to an origin, or to one another.
We estimate that METR’s report implies a 1.7x AI-attributable acceleration in the pace of progress today, relative to historical levels.
Derived from METR's ~2x total acceleration and its guess that ~80% of the acceleration is due to AI; the authors give 1.67–1.74x depending on reading.
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="72048c9d-d8b0-458a-86fc-eaae8f83193a"><a href="https://minerval.ai/claims/72048c9d-d8b0-458a-86fc-eaae8f83193a">Anthropic's AI-attributable AI-capability progress rate in November 2026, relative to its pre-October 2025 rate</a></blockquote> <script async src="https://minerval.ai/embed.js"></script>
<iframe src="https://minerval.ai/embed/claims/72048c9d-d8b0-458a-86fc-eaae8f83193a" title="Minerval: Anthropic's AI-attributable AI-capability progress rate in November 2026, relative to its…" width="480" height="275" style="border:0;max-width:100%" loading="lazy"></iframe>
[](https://minerval.ai/claims/72048c9d-d8b0-458a-86fc-eaae8f83193a)
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 14, 2026. Every judgment on this page is accompanied by a reasoning trace.