BioSimBench 1.0: Gene knockdown
Science
Perturbation-specific F1 on 60 held-out CRISPRi cases predicting which other genes respond when a held-out gene is knocked down.
Benchmark source: Celltype
Top models (higher is better)
| Model | Score | Source |
|---|---|---|
| Gemini 4 Argon | 15.4 | BioSimBench 1.0: Gemini 4 Argon · CellType |
| Gemini 3.1 Pro Preview | 13.1 | BioSimBench 1.0: Gemini 3.1 Pro · CellType |
| GPT-6 Astra | 13.0 | BioSimBench 1.0: GPT-6 Astra · CellType |
| GPT-6.1 Sol | 12.9 | BioSimBench 1.0: GPT-6.1 Sol · CellType |
| GPT-6 Sol | 12.9 | BioSimBench 1.0: GPT-6 Sol · CellType |
| Claude Haiku 5.5 | 12.5 | BioSimBench 1.0: Claude Haiku 5.5 · CellType |
| GPT-6 Luna | 12.3 | BioSimBench 1.0: GPT-6 Luna · CellType |
| Grok 4.7 | 12.1 | BioSimBench 1.0: Grok 4.7 · CellType |
| Gemini 3.8 Flash | 11.5 | BioSimBench 1.0: Gemini 3.8 Flash · CellType |
| Claude Sonnet 5.5 | 11.4 | BioSimBench 1.0: Claude Sonnet 5.5 · CellType |
| Claude Opus 5.5 | 9.9 | BioSimBench 1.0: Claude Opus 5.5 · CellType |