PolyChartQA: leaderboard

Question answering over multi-chart images on PolyChartQA (2026, Beyond Single Plots): 519 human-written questions about 237 figures that combine several related charts from computer science papers, requiring data retrieval, structural understanding and cross-chart reasoning. Zero-shot prompting; accuracy in percent rated by human evaluators, higher is better.

Metric: H-Accuracy (%, times 100), zero-shot prompting, on the 519 human-authored questions of PolyChartQA (534 multi-chart images with 2,297 sub-charts from computer science papers), answers rated correct by human evaluation (H-Accuracy); higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 9 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Gemini 2.0 Flash70.52#731
2Claude 3.7 Sonnet65.32#455
3GPT-4.161.08#506
4Pixtral-12B53.18#2611
5Gemma 3 4B (IT)21.19#4227
6ChartGemma19.27#8660
7Llama 3.2 11B Vision15.22#4281
8LLaVA-v1.5-7B12.33#7642
9MatCha9.06

No result here: #2 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=polychartqa · How It Works · Data refreshed daily, snapshot 2026-11-22.