SupraBench - Binding Affinity MAE: leaderboard
SupraBench (2026) tests LLMs on supramolecular host-guest chemistry built with domain experts. This board is the mean absolute error of the predicted binding affinity (log Ka) under standard aqueous conditions, zero-shot; lower is better.
Metric: Mean absolute error of the predicted log Ka of host-guest binding under standard aqueous conditions (2,609 records), zero-shot direct prompt; lower is better. Source: arxiv.org. Saturation forecast: Around May 2027. 8 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Flash | 1.25 | #164 |
| 2 | GPT-5.4 Mini | 1.55 | #389 |
| 3 | Llama 3.1 70B Instruct | 1.63 | #1546 |
| 4 | GPT-5.4 Nano | 1.64 | #635 |
| 5 | Qwen 3.5 27B | 1.8 | #359 |
| 6 | Qwen 3.5 9B | 2.49 | #803 |
| 7 | Llama 3.1 8B Instruct | 2.7 | #4246 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=suprabench-binding-affinity-mae · How It Works · Data refreshed daily, snapshot 2026-11-13.