Benchmark RadarChartography
Chartography evaluates vision-language models on 100 chart-understanding tasks sourced from professional practice, with questions authored by domain experts and triple-verified. The benchmark uses pass@1 accuracy as the scoring metric across 30 frontier model configurations.
- Released
- 2026-08-11
- Readiness
- Paper only
- Primary field
- Industrial & Engineering
Why it matters
Existing chart benchmarks are saturated and skewed toward simple formats, leaving a gap in measuring performance on realistic, domain-specific charts. Chartography provides a harder, expert-validated test that can differentiate model capabilities in high-stakes professional contexts.
Motivation
Professionals across medicine, engineering, finance, manufacturing, and the sciences often make consequential decisions from charts.
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.