Chart Reasoning on CHARXIV reasoning
64.2Accuracy (All)HierVA
Evaluation results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| HierVAImage Tools=zoom, code, Skills=dynamic2026.05 | 64.2 | 66.3 | 54.5 | 78.4 | 70.1 | 59.9 | 55 | 2,568.38 | |
| CoT-PlanImage Tools=—, Skills=—2026.05 | 62.4 | 64.8 | 54.1 | 78.2 | 69.4 | 55.1 | 52.8 | 1,947.21 | |
| CoTImage Tools=—, Skills=—2026.05 | 62.1 | 64.2 | 53.5 | 78.9 | 69.1 | 54.4 | 52.5 | 1,926.19 | |
| Thinking w/ ImagesImage Tools=zoom, Skills=—2026.05 | 58.9 | 63.5 | 52.1 | 77.5 | 70.2 | 45.1 | 43 | 2,995.02 | |
| Thinking w/ ImagesImage Tools=zoom, code, Skills=fixed2026.05 | 55.3 | 58.2 | 48.5 | 76.8 | 66.5 | 42.1 | 39.7 | 2,837.14 | |
| Thinking w/ ImagesImage Tools=zoom, code, Skills=—2026.05 | 50.1 | 48.5 | 44.2 | 72.5 | 63.1 | 38.5 | 33.8 | 2,990.23 | |
| DirectImage Tools=—, Skills=—2026.05 | 45.7 | 38.5 | 42.1 | 75.4 | 62.3 | 28.9 | 27 | 702.48 |