197 nations rated against 288 national capabilities — 56,736 rated cells — across the nine strategic domains of the National Capability Framework. Every published capability score is one integer out of 20; each level above it is an unweighted mean of the level below, so hard, soft and economic power each count for a third.
National Capability Ratings
197 National Assessments for 288 National Capabilities
2026 National Capability Ratings by Nation
Every nation's index score, its Pareto domination tier, then its three dimensions and the nine capability domains, all out of 20. Filter to a region, or sort any column; the rank re-numbers with the active filter. The Middle East is filtered in its own right rather than inside Asia Pacific, on the UN's Western Asia boundary — the same one thenation directory groups by.
Tier is the Pareto frontier over the nine domain scores: a nation is in tier 1 when no other nation beats it on all nine at once. It reads a nation's shape rather than its size, so it is not the index in another form: China sits seventh on the index and outside the frontier, while Switzerland and Finland reach it from thirteenth and fifteenth. Sorting on it puts tier 1 first and orders the index inside a tier, which is only somewhere for the rows to sit — a tier has no internal order.How this squares with the Pareto tiers in the analysis. No rating bands appear here: the seven bands are published at capability and capability group level, where a score still answers to the ladder it was judged against. Above that it is a mean of means. Read small gaps with care — the index carries a median coder standard deviation of 0.124, so the order inside a close run is not something the method can resolve.
288 National Capabilities in 73 Capability Groups and 9 Capability Domains
2026 National Capability Ratings by Capability Group
Every component of the index — the nine capability domains and the 73 capability groups inside them — with capabilities counted, the mean across the nations in view, the spread of those scores and the nation that tops them. On each bar the light band is the full range from lowest to highest, the solid band is the middle half from the first to the third quartile, and the white dot is the median. Pick a region and every row is recomputed over that region's nations alone, best performer included.
These are simulated expert codings produced under the GINC protocol by a panel of five language models from four vendors and three jurisdictions, in two rounds, with no human adjudication and no source documents attached to any cell. Round-1 spread, not round-3 agreement, is the honest reliability estimate. The method, the panel, the aggregation rule and the known problems are written up in full in analysis/README.md.