HID-Chart - Chemistry: leaderboard
Metric: Accuracy (%; GPT-5 mini judge, 108 chemistry questions). Source: arxiv.org. Saturation forecast: Estimated already saturated. 16 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-4.1 | 65.7 |
| 2 | GPT-4o | 54.6 |
| 3 | Gemini 2.0 Flash | 54.6 |
| 4 | InternVL2.5-78B | 49.1 |
| 5 | Qwen 2.5 VL 7B | 49.1 |
| 6 | InternVL3-8B | 46.3 |
| 7 | Gemma 3 27B | 38.9 |
Interactive version: theaggregate.ai/benchmark?slug=hid-chart-chemistry · How It Works · Data refreshed daily, snapshot 2026-09-26.