ChartArena - Bar Chart (English): leaderboard
Metric: mAP-high (%) on English bar charts, 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 November 2027. 44 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 57.9 |
| 2 | Qwen 3.5 35B A3B (Thinking) | 46.2 |
| 3 | Gemini 2.5 Pro | 46 |
| 4 | Kimi K2.5 (Non-reasoning) | 45.2 |
| 5 | Gemma 4 31B (IT) | 43.7 |
| 6 | Seed 2.0 Pro (Non-reasoning) | 40.3 |
| 7 | Qwen 3 VL 235B A22B Instruct | 38.4 |
| 8 | GPT-5 | 35.1 |
| 9 | GLM-4.5V | 33.5 |
| 10 | Qwen 3.5 9B | 32.5 |
| 11 | Qwen 3 VL 8B (Thinking) | 32.1 |
| 12 | MiMo-V2-Omni | 31.1 |
| 13 | Qwen 3 VL 8B Instruct | 27.5 |
| 14 | Qwen 2.5 VL 32B Instruct | 27.4 |
| 15 | Qwen 2.5 VL 72B Instruct | 27.1 |
Interactive version: theaggregate.ai/benchmark?slug=chartarena-bar-chart-english · How It Works · Data refreshed daily, snapshot 2026-09-29.