What are the best frontend visualization libraries?

The question

what are the best frontend visualization libraries?

Why it matters

Opened from the research queue (Paul via Leo). Research only: which libraries are strongest for which jobs in 2026, with cited tradeoffs. Not a product and not a pick-a-stack-and-build.

What we currently believe

Fetched official pages do not name one best library. They split the job: d3 for bespoke scenes, observable-plot / vega-lite for grammars, chart-js / apache-echarts / highcharts / plotly-js for template charts, visx vs recharts / nivo for React, deck-gl for GPU maps, cytoscape-js for networks. See frontend-viz-job-split. From 2026-08-21-autoresearch-best-frontend-visualization-libraries.

Evidence we have

• From 2026-08-21-autoresearch-best-frontend-visualization-libraries: D3 “is not a charting library”; Plot is “more efficient” and “necessarily less expressive”; Vega-Lite is a JSON grammar that compiles to Vega.
• From 2026-08-21-autoresearch-best-frontend-visualization-libraries: Chart.js — 8 Canvas types; 1M-point decimation demo; canvas “will not be accessible to screen readers” unless the integrator adds ARIA/fallback.
• From 2026-08-21-autoresearch-best-frontend-visualization-libraries: ECharts — 20+ types; Canvas/SVG; official “10 million data in realtime”; homepage a11y blurb (handbook not fetched).
• From 2026-08-21-autoresearch-best-frontend-visualization-libraries: Highcharts — Core 366 USD/seat; Internal / SaaS / OEM; WCAG 2.2 accessibility module. Grokipedia “$176/year” contradicts the shop.
• From 2026-08-21-autoresearch-best-frontend-visualization-libraries: Plotly.js — MIT; statistical / 3D / map / finance catalog.
• From 2026-08-21-autoresearch-best-frontend-visualization-libraries: visx — D3 math + React DOM; MIT; React 18/19. Recharts — React SVG + D3 submodules. nivo — D3 + React component catalog.
• From 2026-08-21-autoresearch-best-frontend-visualization-libraries: deck.gl — WebGPU/WebGL2 layers; MapLibre/Mapbox/Google/ArcGIS; luma.gl not Three.js. Cytoscape.js — network viz + analysis; MIT.
• No fetched official page crowns a winner. Vendor 2026 listicles were not used as evidence.

Evidence we need

• Official Victory, sigma.js, and vis-network pages (requested; not fetched this pass).
• A shared performance bake-off that official homepages do not provide (Grokipedia’s point-count table conflicts with Chart.js / ECharts homepage numbers).
• ECharts handbook aria page (opt-in vs default).
• SPDX files for Chart.js / ECharts / Plot / Vega-Lite / Recharts / nivo / deck.gl (only visx, Plotly.js, and Cytoscape.js MIT were fetched as license facts).

How to resolve

Do not invent a winner while sources compare jobs. A later pass can fill the unfetched official pages and any independent benchmark that names its method. Leave this question open.

Related

• Queue entry: frontend-visualization-libraries
• frontend-viz-job-split
• visualization-grammar
• batteries-included-chart-library
• react-visualization-layer
• frontend-chart-accessibility
• 2026-08-21-autoresearch-best-frontend-visualization-libraries