Skip to main content

Decomposing the Epoch Capabilities Index

ECI visualized as a weighted average of benchmarks

Published

November 17, 2026

Updated

November 18, 2026

The Epoch Capabilities Index (ECI) puts 257 AI models on a single scale. What differences in that number ultimately mean boil down to scores on 57 different benchmarks. Figure 1 below shows, for any pair of frontier or near-frontier models, how benchmark scores determine their ECIs.

The math underlying the ECI

Very briefly, an AI model’s score on a given benchmark is modeled as a function of its capability \(C_m\) (also referred as its ECI): \[ y_{mb} = S\big(\alpha_b \, [C_m - D_b]\big) \tag{1}\]

where \(D_b\) is the benchmark difficulty, \(\alpha_b\) the benchmark slope, and \(S(x) = 1 / (1 + e^{-x})\) is the S-shaped logistic curve rising from 0 to 1. The scatter of the “noise” in each benchmark’s scores is denoted \(\sigma_b\). Epoch fits all the capabilities, difficulties, slopes, and \(\sigma_b\)’s at once as a Bayesian model.

Interpreting a model’s ECI via its scores

Because all parameters are fit jointly, each benchmark score influences the difficulty, slope, and scatter parameters at the same time as it influences a model’s ECI, \(C_m\). As an approximation, though, we can use the estimated values of each benchmark’s difficulty, slope, and \(\sigma_b\) to assess the influence of a given benchmark score on \(C_m\). Each individual benchmark score can then be thought of as a separate measurement of capability: a score \(y_{mb}\) implies \(\hat{C}_{mb}\), the capability at which that score is most likely.1 A model’s ECI can be thought of as a weighted average of these measurements, \(C_m \approx \sum_b w_{mb} \hat{C}_{mb} / \sum_b w_{mb}\), with weights that are, approximately,2

\[ w_{mb} \propto \frac{\alpha_b^2 \, \hat{p}_{mb} (1 - \hat{p}_{mb})}{\sigma_b^2}, \qquad \hat{p}_{mb} = S\big(\alpha_b \, [C_m - D_b]\big) \tag{2}\]

A benchmark score counts most when the corresponding part of that benchmark’s curve is steep and there is little noise. A score near the top or bottom ends of a benchmark, or one anywhere on a benchmark whose estimated curve is flat, counts for little. Otherwise, the scatter \(\sigma_b\) of benchmark noise runs from 0.040 (METR Time Horizons) to 0.181 (MirrorCode), so, other things equal, a score on the former counts about 21 times as much as one on the latter.

(Figure 1 can optionally display analogous measures for an alternative ECI fit, XCI v0.1 . It is the XB+ variant with the addition of benchmark-specific capability noise, \(\tau_b\).)

Figure 1: ECIs as weighted averages of benchmarks. Dots are the ECIs individual benchmark scores imply, the bars are the weights, and a dashed line shows each model’s ECI. Hover over or tap a dot for the score behind it.

The effective number of benchmarks behind a score can be summarized as \(1 / \sum_b (\text{share}_{mb})^2\): about 7 for Claude 3.5 Sonnet and 10 for Claude Opus 5.5, whose top four benchmarks (WeirdML v3, Remote Labor Index, FrontierSWE, EBR-bench) carry 55% of the weight.

Data: Epoch AI Benchmarking Hub (CC BY 4.0), as of November 17, 2026.

Back to top

Footnotes

  1. For mid-range scores this is close to inverting Equation 1, \(D_b + S^{-1}(y_{mb}) / \alpha_b\) with \(S^{-1}(s) = \ln[s / (1 - s)]\). Near 0% or 100% it can differ by a few ECI points, because the assumed noise there is lopsided.↩︎

  2. Equation 2 approximates the information a score carries about \(C_m\) under Epoch’s likelihood, \([\alpha_b \hat{p}_{mb} (1 - \hat{p}_{mb})]^2 \kappa_b^2 [\psi'(\hat{p}_{mb} \kappa_b) + \psi'((1 - \hat{p}_{mb}) \kappa_b)]\) with \(\kappa_b = 1 / (4 \sigma_b^2) - 1\) and \(\psi'\) the trigamma function. The weight shares shown in Figure 1 use these more exact weights, which make the weighted average the capability that best fits the model’s scores given every benchmark’s fitted curve and scatter. Not included in the calculations depicted in Figure 1 is the way Epoch’s ECI also averages over uncertainty in the benchmark curves, so the two differ slightly: by 0.04 points for the median model at ECI 150 and above, and at most 0.16. (Epoch does not publish \(\sigma_b\) updates, so the values here come from a refit of Epoch’s published model code, which reproduces its ECI to within sampling error.)↩︎