CharXiv-D: leaderboard
Descriptive split of CharXiv (Princeton, 2024): 4 short-answer questions per chart on titles, axis labels, legends and tick values over 2,323 arXiv charts; GPT-4o 84.45%, humans 92.1%.
Metric: Score (%). Source: arxiv.org. Source: benchmarklist.com. Saturation forecast: Estimated already saturated. 92 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | O3 (High) | 95 | #124 (O3) |
| 2 | O4 Mini (High) | 94.3 | #180 (O4 Mini) |
| 3 | GPT-4.5 | 90 | #274 |
| 4 | O1 (High) | 88.9 | #183 (O1) |
| 5 | GPT-4.1 Mini | 88.4 | #355 |
| 6 | GPT-4.1 | 87.9 | #248 |
| 7 | O1 | 87.45 | #183 |
| 8 | InternVL3-38B | 87.2 | #405 |
| 9 | InternVL3-78B | 85.1 | #352 |
| 10 | Claude 3.5 Sonnet | 84.3 | #350 |
| 11 | Doubao-1.5-Pro | 84.3 | #287 |
| 12 | InternVL3-14B | 82.2 | #505 |
| 13 | Qwen 2 VL 72B | 81.35 | #460 |
| 14 | GPT-4o Mini | 74.92 | #599 |
| 15 | GPT-4.1 Nano | 73.9 | #731 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=charxiv-d · How It Works · Data refreshed daily, snapshot 2026-10-20.