Compare
Twenty libraries. One ledger.
ChartKit scored against the 20 libraries in features.json, on the same feature matrix we use to track our own gaps. Pick yours in the table below for the row-by-row.
features.json generated 2026-10-19. Other libraries checked 2026-10-18 against types and schemas. ChartKit rechecked 2026-11-04 on 236653ac.
Ledger
Find the one you use.
Each row opens the pairwise page: every scored feature, both options, the check and the source. A loss is a feature the other library scores higher on. Unknown counts as neither.
Call
Use it when. Skip it when.
The feature count tells you breadth. These are the things that decide it, with the measured numbers.
React, Vue, Svelte, Solid, Qwik, Angular, Astro, vanilla, and the custom element read the same JSON. The server renders it too.
With llms.txt in the prompt, round 6 was 42/42 on the first try. validate() and repair() do not add a model call.
A 100,000-point line server-renders in 2.63 ms median. Recharts 3 SSR returns no SVG except sunburst.
The line chart is 32,110 bytes gzip for createRenderer. Bar, mini-area and /all are on Benchmarks.
Tooltip, zoom, pan, brush and animation live in the client runtime. The static SVG does not paint them.
ChartKit paints the first hover on a 100,000 point line in 2.1 and 2.2 ms. uPlot does it in 1.0 and 1.2 ms (50 and 200 ms after mount). uPlot places no tooltip.
Even 1,000-point update: ChartKit 1.5 ms, uPlot 1.2 ms. Round 13, ChartKit vs uPlot.
Choropleth and point maps are in the spec. There is no tile layer, and no pan or zoom on the map.
Convert
Bring the chart you already have.
Paste a Recharts component, a Chart.js config or an ECharts option and get the ChartSpec back. Chart.js, Recharts and ECharts today.