STEMVerse - Physics - Thermodynamics and Statistical Physics: leaderboard
STEMVerse physics problems labelled with the sub-discipline 'Thermodynamics and Statistical Physics', drawn from existing STEM benchmarks pooled into one set; labels come from GPT-4o and were checked by an expert audit. Few-shot prompting in the MegaScience style; accuracy in percent, higher is better. Li et al. (ZGCA and NTU), 2026 preprint.
Metric: Few-shot accuracy (%). Source: arxiv.org. Saturation forecast: Estimated already saturated. 16 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen3-14B-Base | 28.45 | #1608 |
| 2 | Qwen3-8B-Base | 20.22 | #1728 |
| 3 | Llama 3.1 8B Instruct | 20.08 | #4277 |
| 4 | Llama 3.1 8B | 17.99 | #4900 |
| 5 | Qwen 2.5 3B | 16.32 | #5214 |
| 6 | Llama 3.2 3B | 15.48 | #7877 |
| 7 | Qwen 2.5 14B Instruct | 15.06 | #1745 |
| 8 | Qwen 2.5 7B | 15.06 | #2718 |
| 9 | Qwen 2.5 14B | 14.37 | #1794 |
| 10 | Llama 3.2 3B Instruct | 12.13 | #7185 |
| 11 | Qwen3-4B-Base | 11.58 | #2444 |
| 12 | Qwen 2.5 7B Instruct | 10.46 | #2583 |
| 13 | Qwen 2.5 3B Instruct | 9.92 | #5206 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=stemverse-physics-thermodynamics-and-statistical-physics · How It Works · Data refreshed daily, snapshot 2026-11-12.