SupraBench - Host-Guest Description (Few-Shot): leaderboard
Open-ended host-guest description in SupraBench (2026), a supramolecular chemistry benchmark: ROUGE-1 F1 times 100 of generated descriptions of candidate hosts or guests against expert references, with three to five in-domain demonstrations. Higher is better.
Metric: ROUGE-1 F1 x100 (0-100) of open-ended descriptions of candidate hosts or binding guests against expert references (135 questions), three to five in-domain demonstrations. Source: arxiv.org. Saturation forecast: Around 2031. 8 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3.5 27B | 58 | #359 |
| 2 | Llama 3.1 70B Instruct | 53.1 | #1546 |
| 3 | Llama 3.1 8B Instruct | 45.6 | #4246 |
| 4 | Gemini 3 Flash | 44.8 | #164 |
| 5 | GPT-5.4 Mini | 30.7 | #389 |
| 6 | GPT-5.4 Nano | 15.2 | #635 |
| 7 | Qwen 3.5 9B | 4.2 | #803 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=suprabench-host-guest-description-few-shot · How It Works · Data refreshed daily, snapshot 2026-11-13.