RefusalBench: leaderboard
Frontier large language models are increasingly deployed as orchestration backbones for biological research workflows, yet no shared evidence base exists for comparing their refusal behaviour on legitimate research prompts. RefusalBench, introduced here, is a matched-triple benchmark of 141 prompts in 47 bundles that holds task framing constant while varying only biological risk tier (benign, bord
Metric: Strict Refusal Rate (self-reported). Source: benchmarklist.com. Saturation forecast: Estimated already saturated. 19 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.7 | 0.1 | #49 |
| 2 | GPT-4.1 Nano | 0.2 | #731 |
| 3 | Gemini 2.5 Pro | 0.4 | #152 |
| 4 | Gemini 2.5 Flash | 0.6 | #245 |
| 5 | GPT-4.1 Mini | 0.6 | #355 |
| 6 | Claude Opus 4.5 | 1.2 | #83 |
| 7 | Grok 4.20 | 1.3 | #175 |
| 8 | GPT-4.1 | 1.4 | #248 |
| 9 | GPT-4o | 2.1 | #343 |
| 10 | DeepSeek R1 | 18.4 | #251 |
| 11 | Qwen 2.5 Max | 29.8 | #416 |
| 12 | DeepSeek V3 (0324) | 47.5 | #340 |
| 13 | Mistral Large 2 (Jul) | 87.2 | #484 |
| 14 | Command A | 94.6 | #508 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=refusalbench · How It Works · Data refreshed daily, snapshot 2026-10-20.