ChartArena - Flowchart (English): leaderboard

Metric: mAP-high (%) on English flowcharts, 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 December 2026. 40 models tracked.

Top models

#ModelScore
1Gemini 3.1 Pro (Preview)65.6
2Seed 2.0 Pro (Non-reasoning)62.6
3Qwen 3.5 35B A3B (Thinking)62.5
4Gemini 2.5 Pro62.1
5Kimi K2.5 (Non-reasoning)59.9
6Gemma 4 31B (IT)59.6
7GPT-558.1
8Qwen 3 VL 235B A22B Instruct57.9
9MiMo-V2-Omni57.1
10Qwen 2.5 VL 72B Instruct50.1
11Qwen 3 VL 8B Instruct50
12GPT-4o49.8
13Qwen 3 VL 8B (Thinking)45.7
14Qwen 3.5 9B45.5
15Qwen 3 VL 4B Instruct44.7

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