Nutrient Chart Parsing: leaderboard
Nutrient Chart Parsing Leaderboard: models extract the underlying data from 2,500 charts. Headline mean per-chart cell F1 at 3% tolerance (exact when the chart prints data labels, 3 percentage points for share charts, otherwise 3% of the value range); %, higher is better. The publisher's own commercial chart-parsing model is ranked alongside provider APIs and open-weight VLMs.
Metric: Mean per-chart cell F1 at 3% tolerance (%, 2,500 charts). Source: huggingface.co. Saturation forecast: Around May 2027. 13 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.8 Flash | 71.02 | #30 |
| 2 | Claude Opus 5 | 69.71 | #16 |
| 3 | GPT-6 Astra | 67.47 | #6 |
| 4 | Gemini 3.5 Flash Lite | 66.35 | #174 |
| 5 | Claude Sonnet 5 | 65.7 | #65 |
| 6 | GPT-5.6 Terra | 62.96 | #40 |
| 7 | Gemini 3.1 Pro (Preview) | 62.81 | #64 |
| 8 | Claude Haiku 4.5 (20251001) | 57.05 | #264 |
| 9 | GPT-5.6 Sol | 56.59 | #22 |
| 10 | Qwen 3.8 27B | 52.35 | #123 |
| 11 | Qwen 3 VL 4B Instruct | 51.81 | #510 |
| 12 | Qwen 3.5 4B | 16.87 | #514 |
No result here: #3 Claude Opus 5.5, #7 GPT-6.1 Sol, #9 Gemini 4 Argon.
Interactive version: theaggregate.ai/benchmark?slug=nutrient-chart-parsing · How It Works · Data refreshed daily, snapshot 2026-11-05.