Benchmarks.bio - BioSecBench-Refusal: leaderboard
BioSecBench-Refusal (benchmarks.bio) — biosecurity refusal and over-refusal benchmark; red-team prompts that should be refused plus routine biology work that should be answered, scored as combined pass rate.
Metric: Pass Rate (%). Source: benchmarks.bio. Saturation forecast: Not forecast. 19 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Grok 4.6 | 64.2 | #26 |
| 2 | Gemini 3.5 Flash | 51.5 | #60 |
| 3 | GPT-5.4 | 45.7 | #79 |
| 4 | GPT-6 Astra | 44.6 | #6 |
| 5 | Claude Sonnet 4.6 | 43.5 | #90 |
| 6 | Gemini 3.1 Pro (Preview) | 43 | #58 |
| 7 | GPT-5.6 Sol | 39.5 | #19 |
| 8 | Claude Opus 4.6 | 39 | #63 |
| 9 | GPT-5.5 | 37.1 | #31 |
| 10 | Claude Opus 4.7 | 36.7 | #49 |
| 11 | GPT-5.6 Terra | 34.3 | #41 |
| 12 | GPT-5.6 Luna | 33.3 | #68 |
| 13 | Claude Sonnet 5 | 30.1 | #57 |
| 14 | Claude Opus 5 | 29.4 | #12 |
| 15 | Claude Opus 4.8 | 28.6 | #37 |
No result here: #3 Claude Opus 5.5, #7 GPT-6.1 Sol, #8 Gemini 4 Argon.
Interactive version: theaggregate.ai/benchmark?slug=benchmarks-bio-biosecbench-refusal · How It Works · Data refreshed daily, snapshot 2026-10-20.