Family
Benchmark · 2026
Chartography
A deliberately difficult set of practitioner-authored professional chart-reading tasks.
Open the primary source ↗Task
Solve 100 tasks across 12 domain labels, repeated across 30 model configurations.
Evidence
Mean pass at one over 20 runs per task, with three expert verifiers and acceptable answer ranges.
What it establishes
Makes difficult professional chart reading a distinct capability and shows that more reasoning often helps only modestly.
What it does not establish
Difficulty screening means the score is not the failure rate on ordinary professional charts.
Related entities
Continue through the library.
- ChartQAProA realistic chart question-answering benchmark with conversational, hypothetical, fact-checking, and unanswerable cases.
- PolyChartQA: multi-chart figuresA multi-chart scientific-figure benchmark whose name collides with a separate multilingual chart benchmark.
- MM-JudgeBenchA multilingual pairwise-preference benchmark for multimodal judges that includes a chart-centric subset.