EpiBench: leaderboard
Task 2 of the epitope-centered EpiBench: identify which candidate antigen region a given antibody binds. Reports zero-shot epitope-identification accuracy from Table 2, not an aggregate of the paper's five biological tasks.
Metric: Epitope Identification Accuracy (%, Task 2). Source: arxiv.org. Saturation forecast: Not forecast. 9 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Grok 4.20 | 33.6 | #175 |
| 2 | GPT-5.5 | 31.9 | #31 |
| 3 | DeepSeek V4 Pro | 30.8 | #102 |
| 4 | Qwen 3 235B A22B | 29.4 | #313 |
| 5 | Gemini 3 Flash | 28.2 | #99 |
| 6 | GPT-4o | 25.4 | #343 |
| 7 | Kimi K2.6 | 25.4 | #104 |
| 8 | GLM-5.2 | 24.3 | #73 |
| 9 | Qwen 3.7 Plus | 24 | #92 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=epibench · How It Works · Data refreshed daily, snapshot 2026-10-20.