STEMVerse - Chemistry - Chemical Engineering and Technology: leaderboard
Metric: Few-shot accuracy (%). Source: arxiv.org. Saturation forecast: Estimated already saturated. 16 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen3-4B-Base | 34.04 |
| 2 | Qwen3-14B-Base | 30.85 |
| 3 | Qwen 2.5 14B Instruct | 24.47 |
| 4 | Qwen 2.5 7B | 22.34 |
| 5 | Qwen3-8B-Base | 22.34 |
| 6 | Qwen 2.5 14B | 20.21 |
| 7 | Qwen 2.5 7B Instruct | 18.09 |
| 8 | Qwen 2.5 3B Instruct | 18.09 |
| 9 | Llama 3.1 8B Instruct | 15.96 |
| 10 | Llama 3.1 8B | 14.89 |
| 11 | Qwen 2.5 3B | 14.89 |
| 12 | Llama 3.2 3B Instruct | 10.64 |
| 13 | Llama 3.2 3B | 4.26 |
Interactive version: theaggregate.ai/benchmark?slug=stemverse-chemistry-chemical-engineering-and-technology · How It Works · Data refreshed daily, snapshot 2026-09-25.