SupraBench - Host-Guest Description: leaderboard

Metric: ROUGE-1 F1 x100 (0-100) of open-ended descriptions of candidate hosts or binding guests against expert references (135 questions), zero-shot direct prompt. Source: arxiv.org. Saturation forecast: Around 2034. 8 models tracked.

Top models

#ModelScore
1Llama 3.1 70B Instruct15.2
2GPT-5.4 Mini13.7
3Qwen 3.5 27B12.2
4Gemini 3 Flash11.8
5GPT-5.4 Nano10.7
6Llama 3.1 8B Instruct9.2
7Qwen 3.5 9B4.3

Interactive version: theaggregate.ai/benchmark?slug=suprabench-host-guest-description · How It Works · Data refreshed daily, snapshot 2026-09-29.