ChartArena - Box Plot (English): leaderboard

Metric: mAP-high (%) on English box plots, each model parses the chart image into its native structured format, normalized format-agnostically to (header, entity, value) triples for numeric charts or to labeled graphs and trees for flowcharts and mind maps; mean average precision over similarity thresholds at the high tolerance level, averaged over three visual styles (digital renderings, printed photos, hand-drawn photos); higher is better. Source: arxiv.org. Saturation forecast: Around July 2027. 44 models tracked.

Top models

#ModelScore
1Seed 2.0 Pro (Non-reasoning)44.5
2Qwen 3.5 35B A3B (Thinking)42.2
3Kimi K2.5 (Non-reasoning)40.6
4Gemini 3.1 Pro (Preview)32.5
5GLM-4.5V32.4
6Qwen 3 VL 8B (Thinking)22.4
7Qwen 3 VL 4B (Thinking)22.1
8GPT-519.8
9GPT-4o19.1
10MiMo-V2-Omni19.1
11Qwen 3.5 9B15.3
12Qwen 3 VL 4B Instruct15.3
13Qwen 2.5 VL 72B Instruct15
14Qwen 3 VL 235B A22B Instruct14.1
15Qwen 3 VL 8B Instruct11.6

Interactive version: theaggregate.ai/benchmark?slug=chartarena-box-plot-english · How It Works · Data refreshed daily, snapshot 2026-09-29.