BioTIER Refuse
The official leaderboard hasn’t scored a model released after GPT-6.1 Terra (October 2026). Labs report their own numbers for one newer model: GPT-6.2 Astra at 79.8%.
Models receive 398 dangerous biology prompts. The score measures the percentage correctly refused.
One track of BioTIER. The BioTIER record keeps lab scores that combine tracks.
- Made by
- SecureBio
- Released
- March 2026
- Kind
- Refusing harmful requests
- Level
- Expert
- Scored
- Graded by a model
- Answers
- Open-ended
- Tasks
- Kept private
- Measures
- Propensity
- Behavior
- Harmful use
- Subject
- Biological sciences
- Language
- English
- Interaction
- Single turn
- Items from
- Another benchmark
- Refresh
- One fixed set
- Run by
- A host
- Based on
- BioTIER
- In views
- Biology and medicine, Content safety
- Strongest model tested
- GPT-6.2 Astra #1 on the capability index
- Last in a system card
- November 2026
- Last model added
- November 2026
- Scores refreshed
- November 18th, 2026
Claude Opus 5.5 leads with 99.1% Claude Opus 5.5 leads with 99.1%
Best score per entry from the benchmark's own leaderboard, read on November 18th, 2026. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. This view shows only models we track, each at its best.
Claude Opus 5.5 leads with 99.1%, claims included
Best score per model from every source, including labs’ own numbers and claims Epoch AI collected from reports and papers. Labs pick their own settings, so a claim can beat an independent run of the same model. Striped bars are claims, and dotted bars are reruns by Epoch AI or another board.
- 99.1 Claude Opus 5.5
- 98.2 Claude Sonnet 4.5
- 97.9 Claude Haiku 5.5
- 96.1 Claude Sonnet 5
- 95.9 Claude Opus 4.8
- 95.7 Claude Sonnet 5.5
- 95.5
Muse Spark Open
- 95.2 Claude Sonnet 4.6
- 94.9 Claude Opus 4.7
- 94.9 Claude Opus 5
- 94.4 Claude Opus 4.6
- 92.6 Claude Opus 4.5
- 87.8
Gemini 4 Argon
- 81.7
Gemini 3.5 Flash
- 81.1
Gemini 3.6 Flash
- 80.3 Grok 4.7
- 79.8 GPT-6 Sol
- 79.5 GPT-6 Luna
- 79.1 GPT-6.1 Terra
- 78.2
Gemini 3.7 Flash
- 78.1 GPT-6 Astra
- 78.0
Gemini 3.8 Flash
- 77.8 GPT-6.1 Sol
- 76.9 GPT-5.4
- 76.3
Muse Glimmer
- 76.2 Grok 4.5
- 76.1 Grok 4.6
- 75.4 GPT-5.2
- 75.3 GPT-5.5
- 74.0 GPT-5.6 Sol
- 73.6 GPT-5.5 Pro
- 72.1 GPT-5.6 Luna
- 71.7 GPT-5.6 Terra
- 68.7
Gemini 3.1 Pro
- 66.9 Grok 4.20
- 66.0
Gemini 3 Flash
- 61.6
Gemma 4 31B IT
- 61.4 GPT-Rosalind-5.5
- 61.3 Beam 1.0
- 45.4 gpt-oss-120b
- 42.0 Claude Haiku 4.5
- 39.3 gpt-oss-20b
- 35.7
Qwen3.5 397B-A17B
- 33.9 GLM-5
- 32.8
Qwen3.7-Max
- 32.6 GLM-5.2
- 32.6 Kimi K3.1 Pro
- 32.5 GLM-5.3
- 32.4 Nemotron 3 Ultra 550B A55B
- 32.2
Mistral Large 4
- 32.1
Nemotron 3.5 Super VL
- 31.6
Qwen3.8-2.4T-A95B
- 30.0
Inkling
- 27.3 Kimi K3
- 25.9
DeepSeek V4.1 Flash 1017
- 24.9
Inkling-Small
- 23.9 Nemotron 3 Super 120B A12B
- 23.7 GLM-5.1
- 22.1 GPT-4o
- 21.9 Kimi K2.6
- 21.9
Qwen 3.8 27B
- 21.2
Llama 3.3 70B
- 21.2 Kimi K2.5
- 21.1
Nemotron 3.5 Lightning 30B-A3B
- 20.9
Qwen3-235B-A22B-Thinking
- 20.2
Nemotron 3 Nano 30B-A3B
- 17.6
DeepSeek V4 Pro 0813
- 16.7 GPT-4.1
- 16.3 Kimi K2 Thinking
- 15.6
DeepSeek-R1
- 15.4 GPT-4o mini (snapshot unspecified)
- 14.5 GPT-4o
- 10.4 GPT-4.1 mini
- 10.3
Mistral Medium 3.1
- 10.2
DeepSeek-V3.2
- 9.30 Gemini 2.5 Flash (snapshot unspecified)
- 9.20
Mistral Large 3
- 1 Claude Opus 5.5 max 99.1%
- 2 Claude Sonnet 4.5 16k 98.2%
- 3 Claude Haiku 5.5 max 97.9%
- 4 Claude Sonnet 5 max 96.1%
- 5 Claude Opus 4.8 max 95.9%
- 6 Claude Sonnet 5.5 max 95.7%
- 7
Muse Spark Open xhigh 95.5%
- 8 Claude Sonnet 4.6 max 95.2%
- 9 Claude Opus 4.7 max 94.9%
- 10 Claude Opus 5 max 94.9%
- 11 Claude Opus 4.6 max 94.4%
- 12 Claude Opus 4.5 16k 92.6%
- 13
Gemini 4 Argon high 87.8%
- 14
Gemini 3.5 Flash high 81.7%
- 15
Gemini 3.6 Flash high 81.1%
- 16 Grok 4.7 xhigh 80.3%
- 17 GPT-6 Sol max 79.8%
- 18 GPT-6 Luna max 79.5%
- 19 GPT-6.1 Terra max 79.1%
- 20
Gemini 3.7 Flash high 78.2%
- 21 GPT-6 Astra max 78.1%
- 22
Gemini 3.8 Flash high 78%
- 23 GPT-6.1 Sol max 77.8%
- 24 GPT-5.4 xhigh 76.9%
- 25
Muse Glimmer xhigh 76.3%
- 26 Grok 4.5 high 76.2%
- 27 Grok 4.6 xhigh 76.1%
- 28 GPT-5.2 xhigh 75.4%
- 29 GPT-5.5 xhigh 75.3%
- 30 GPT-5.6 Sol max 74%
- 31 GPT-5.5 Pro xhigh 73.6%
- 32 GPT-5.6 Luna max 72.1%
- 33 GPT-5.6 Terra max 71.7%
- 34
Gemini 3.1 Pro high 68.7%
- 35 Grok 4.20 reasoning 66.9%
- 36
Gemini 3 Flash high 66%
- 37
Gemma 4 31B IT 61.6%
- 38 GPT-Rosalind-5.5 xhigh 61.4%
- 39 Beam 1.0 max 61.3%
- 40 gpt-oss-120b 45.4%
- 41 Claude Haiku 4.5 16k 42%
- 42 gpt-oss-20b 39.3%
- 43
Qwen3.5 397B-A17B 35.7%
- 44 GLM-5 33.9%
- 45
Qwen3.7-Max 32.8%
- 46 GLM-5.2 max 32.6%
- 47 Kimi K3.1 Pro max 32.6%
- 48 GLM-5.3 max 32.5%
- 49 Nemotron 3 Ultra 550B A55B 32.4%
- 50
Mistral Large 4 high 32.2%
- 51
Nemotron 3.5 Super VL 32.1%
- 52
Qwen3.8-2.4T-A95B xhigh 31.6%
- 53
Inkling xhigh 30%
- 54 Kimi K3 max 27.3%
- 55
DeepSeek V4.1 Flash 1017 max 25.9%
- 56
Inkling-Small xhigh 24.9%
- 57 Nemotron 3 Super 120B A12B 23.9%
- 58 GLM-5.1 23.7%
- 59 GPT-4o Nov 2024 22.1%
- 60 Kimi K2.6 21.9%
- 61
Qwen 3.8 27B xhigh 21.9%
- 62
Llama 3.3 70B 21.2%
- 63 Kimi K2.5 21.2%
- 64
Nemotron 3.5 Lightning 30B-A3B 21.1%
- 65
Qwen3-235B-A22B-Thinking Jul 2025 20.9%
- 66
Nemotron 3 Nano 30B-A3B 20.2%
- 67
DeepSeek V4 Pro 0813 max 17.6%
- 68 GPT-4.1 16.7%
- 69 Kimi K2 Thinking 16.3%
- 70
DeepSeek-R1 May 2025 15.6%
- 71 GPT-4o mini (snapshot unspecified) 15.4%
- 72 GPT-4o Aug 2024 14.5%
- 73 GPT-4.1 mini 10.4%
- 74
Mistral Medium 3.1 10.3%
- 75
DeepSeek-V3.2 10.2%
- 76 Gemini 2.5 Flash (snapshot unspecified) 16k 9.30%
- 77
Mistral Large 3 9.20%
Source: Official BioTIER Refuse leaderboard
- 99.1 Claude Opus 5.5
- 98.2 Claude Sonnet 4.5
- 97.9 Claude Haiku 5.5
- 96.1 Claude Sonnet 5
- 95.9 Claude Opus 4.8
- 95.7 Claude Sonnet 5.5
- 95.5
Muse Spark Open
- 95.2 Claude Sonnet 4.6
- 94.9 Claude Opus 4.7
- 94.9 Claude Opus 5
- 94.4 Claude Opus 4.6
- 92.6 Claude Opus 4.5
- 87.8
Gemini 4 Argon
- 81.7
Gemini 3.5 Flash
- 81.1
Gemini 3.6 Flash
- 80.3 Grok 4.7
- 79.8 GPT-6 Sol
- 79.5 GPT-6 Luna
- 79.1 GPT-6.1 Terra
- 78.2
Gemini 3.7 Flash
- 78.1 GPT-6 Astra
- 78.0
Gemini 3.8 Flash
- 77.8 GPT-6.1 Sol
- 76.9 GPT-5.4
- 76.3
Muse Glimmer
- 76.2 Grok 4.5
- 76.1 Grok 4.6
- 75.4 GPT-5.2
- 75.3 GPT-5.5
- 74.0 GPT-5.6 Sol
- 73.6 GPT-5.5 Pro
- 72.1 GPT-5.6 Luna
- 71.7 GPT-5.6 Terra
- 68.7
Gemini 3.1 Pro
- 66.9 Grok 4.20
- 66.0
Gemini 3 Flash
- 61.6
Gemma 4 31B IT
- 61.4 GPT-Rosalind-5.5
- 61.3 Beam 1.0
- 45.4 gpt-oss-120b
- 42.0 Claude Haiku 4.5
- 39.3 gpt-oss-20b
- 35.7
Qwen3.5 397B-A17B
- 33.9 GLM-5
- 32.8
Qwen3.7-Max
- 32.6 GLM-5.2
- 32.6 Kimi K3.1 Pro
- 32.5 GLM-5.3
- 32.2
Mistral Large 4
- 32.1
Nemotron 3.5 Super VL
- 31.6
Qwen3.8-2.4T-A95B
- 30.0
Inkling
- 27.3 Kimi K3
- 25.9
DeepSeek V4.1 Flash 1017
- 24.9
Inkling-Small
- 23.7 GLM-5.1
- 22.1 GPT-4o
- 21.9 Kimi K2.6
- 21.9
Qwen 3.8 27B
- 21.2
Llama 3.3 70B
- 21.2 Kimi K2.5
- 21.1
Nemotron 3.5 Lightning 30B-A3B
- 20.9
Qwen3-235B-A22B-Thinking
- 20.2
Nemotron 3 Nano 30B-A3B
- 17.6
DeepSeek V4 Pro 0813
- 16.7 GPT-4.1
- 16.3 Kimi K2 Thinking
- 15.6
DeepSeek-R1
- 14.5 GPT-4o
- 10.4 GPT-4.1 mini
- 10.3
Mistral Medium 3.1
- 10.2
DeepSeek-V3.2
- 9.20
Mistral Large 3
- 1 Claude Opus 5.5 max 99.1%
- 2 Claude Sonnet 4.5 16k 98.2%
- 3 Claude Haiku 5.5 max 97.9%
- 4 Claude Sonnet 5 max 96.1%
- 5 Claude Opus 4.8 max 95.9%
- 6 Claude Sonnet 5.5 max 95.7%
- 7
Muse Spark Open xhigh 95.5%
- 8 Claude Sonnet 4.6 max 95.2%
- 9 Claude Opus 4.7 max 94.9%
- 10 Claude Opus 5 max 94.9%
- 11 Claude Opus 4.6 max 94.4%
- 12 Claude Opus 4.5 16k 92.6%
- 13
Gemini 4 Argon high 87.8%
- 14
Gemini 3.5 Flash high 81.7%
- 15
Gemini 3.6 Flash high 81.1%
- 16 Grok 4.7 xhigh 80.3%
- 17 GPT-6 Sol max 79.8%
- 18 GPT-6 Luna max 79.5%
- 19 GPT-6.1 Terra max 79.1%
- 20
Gemini 3.7 Flash high 78.2%
- 21 GPT-6 Astra max 78.1%
- 22
Gemini 3.8 Flash high 78%
- 23 GPT-6.1 Sol max 77.8%
- 24 GPT-5.4 xhigh 76.9%
- 25
Muse Glimmer xhigh 76.3%
- 26 Grok 4.5 high 76.2%
- 27 Grok 4.6 xhigh 76.1%
- 28 GPT-5.2 xhigh 75.4%
- 29 GPT-5.5 xhigh 75.3%
- 30 GPT-5.6 Sol max 74%
- 31 GPT-5.5 Pro xhigh 73.6%
- 32 GPT-5.6 Luna max 72.1%
- 33 GPT-5.6 Terra max 71.7%
- 34
Gemini 3.1 Pro high 68.7%
- 35 Grok 4.20 reasoning 66.9%
- 36
Gemini 3 Flash high 66%
- 37
Gemma 4 31B IT 61.6%
- 38 GPT-Rosalind-5.5 xhigh 61.4%
- 39 Beam 1.0 max 61.3%
- 40 gpt-oss-120b 45.4%
- 41 Claude Haiku 4.5 16k 42%
- 42 gpt-oss-20b 39.3%
- 43
Qwen3.5 397B-A17B 35.7%
- 44 GLM-5 33.9%
- 45
Qwen3.7-Max 32.8%
- 46 GLM-5.2 max 32.6%
- 47 Kimi K3.1 Pro max 32.6%
- 48 GLM-5.3 max 32.5%
- 49
Mistral Large 4 high 32.2%
- 50
Nemotron 3.5 Super VL 32.1%
- 51
Qwen3.8-2.4T-A95B xhigh 31.6%
- 52
Inkling xhigh 30%
- 53 Kimi K3 max 27.3%
- 54
DeepSeek V4.1 Flash 1017 max 25.9%
- 55
Inkling-Small xhigh 24.9%
- 56 GLM-5.1 23.7%
- 57 GPT-4o Nov 2024 22.1%
- 58 Kimi K2.6 21.9%
- 59
Qwen 3.8 27B xhigh 21.9%
- 60
Llama 3.3 70B 21.2%
- 61 Kimi K2.5 21.2%
- 62
Nemotron 3.5 Lightning 30B-A3B 21.1%
- 63
Qwen3-235B-A22B-Thinking Jul 2025 20.9%
- 64
Nemotron 3 Nano 30B-A3B 20.2%
- 65
DeepSeek V4 Pro 0813 max 17.6%
- 66 GPT-4.1 16.7%
- 67 Kimi K2 Thinking 16.3%
- 68
DeepSeek-R1 May 2025 15.6%
- 69 GPT-4o Aug 2024 14.5%
- 70 GPT-4.1 mini 10.4%
- 71
Mistral Medium 3.1 10.3%
- 72
DeepSeek-V3.2 10.2%
- 73
Mistral Large 3 9.20%
Source: Official BioTIER Refuse leaderboard
- 99.1 Claude Opus 5.5
- 98.7
Muse Wonder Full
- 98.2 Claude Sonnet 4.5
- 97.9 Claude Haiku 5.5
- 96.1 Claude Sonnet 5
- 95.9 Claude Opus 4.8
- 95.7 Claude Sonnet 5.5
- 95.5
Muse Spark Open
- 95.2 Claude Sonnet 4.6
- 94.9 Claude Opus 4.7
- 94.9 Claude Opus 5
- 94.4 Claude Opus 4.6
- 92.6 Claude Opus 4.5
- 87.8
Gemini 4 Argon
- 82.9 GPT-6 Astra
- 81.7
Gemini 3.5 Flash
- 81.1
Gemini 3.6 Flash
- 80.3 Grok 4.7
- 79.8 GPT-6.2 Astra
- 79.8 GPT-6 Sol
- 79.5 GPT-6 Luna
- 79.1 GPT-6.1 Terra
- 78.2
Gemini 3.7 Flash
- 78.0
Gemini 3.8 Flash
- 77.8 GPT-6.1 Sol
- 76.9 GPT-5.4
- 76.3
Muse Glimmer
- 76.2 Grok 4.5
- 76.1 Grok 4.6
- 75.4 GPT-5.2
- 75.3 GPT-5.5
- 74.0 GPT-5.6 Sol
- 73.6 GPT-5.5 Pro
- 72.1 GPT-5.6 Luna
- 71.7 GPT-5.6 Terra
- 68.7
Gemini 3.1 Pro
- 66.9 Grok 4.20
- 66.0
Gemini 3 Flash
- 64.6 Beam 1.0
- 61.6
Gemma 4 31B IT
- 61.4 GPT-Rosalind-5.5
- 45.4 gpt-oss-120b
- 42.0 Claude Haiku 4.5
- 39.3 gpt-oss-20b
- 35.7
Qwen3.5 397B-A17B
- 33.9 GLM-5
- 32.8
Qwen3.7-Max
- 32.6 GLM-5.2
- 32.6 Kimi K3.1 Pro
- 32.5 GLM-5.3
- 32.2
Mistral Large 4
- 32.1
Nemotron 3.5 Super VL
- 31.6
Qwen3.8-2.4T-A95B
- 30.0
Inkling
- 27.3 Kimi K3
- 25.9
DeepSeek V4.1 Flash 1017
- 24.9
Inkling-Small
- 23.7 GLM-5.1
- 22.1 GPT-4o
- 21.9 Kimi K2.6
- 21.9
Qwen 3.8 27B
- 21.2
Llama 3.3 70B
- 21.2 Kimi K2.5
- 21.1
Nemotron 3.5 Lightning 30B-A3B
- 20.9
Qwen3-235B-A22B-Thinking
- 20.2
Nemotron 3 Nano 30B-A3B
- 17.6
DeepSeek V4 Pro 0813
- 16.7 GPT-4.1
- 16.3 Kimi K2 Thinking
- 15.6
DeepSeek-R1
- 14.5 GPT-4o
- 10.4 GPT-4.1 mini
- 10.3
Mistral Medium 3.1
- 10.2
DeepSeek-V3.2
- 9.20
Mistral Large 3
- 1 Claude Opus 5.5 max 99.1%
- 2
Muse Wonder Full 98.7%
- 3 Claude Sonnet 4.5 16k 98.2%
- 4 Claude Haiku 5.5 max 97.9%
- 5 Claude Sonnet 5 max 96.1%
- 6 Claude Opus 4.8 max 95.9%
- 7 Claude Sonnet 5.5 max 95.7%
- 8
Muse Spark Open xhigh 95.5%
- 9 Claude Sonnet 4.6 max 95.2%
- 10 Claude Opus 4.7 max 94.9%
- 11 Claude Opus 5 max 94.9%
- 12 Claude Opus 4.6 max 94.4%
- 13 Claude Opus 4.5 16k 92.6%
- 14
Gemini 4 Argon high 87.8%
- 15 GPT-6 Astra 82.9%
- 16
Gemini 3.5 Flash high 81.7%
- 17
Gemini 3.6 Flash high 81.1%
- 18 Grok 4.7 xhigh 80.3%
- 19 GPT-6.2 Astra 79.8%
- 20 GPT-6 Sol max 79.8%
- 21 GPT-6 Luna max 79.5%
- 22 GPT-6.1 Terra max 79.1%
- 23
Gemini 3.7 Flash high 78.2%
- 24
Gemini 3.8 Flash high 78%
- 25 GPT-6.1 Sol max 77.8%
- 26 GPT-5.4 xhigh 76.9%
- 27
Muse Glimmer xhigh 76.3%
- 28 Grok 4.5 high 76.2%
- 29 Grok 4.6 xhigh 76.1%
- 30 GPT-5.2 xhigh 75.4%
- 31 GPT-5.5 xhigh 75.3%
- 32 GPT-5.6 Sol max 74%
- 33 GPT-5.5 Pro xhigh 73.6%
- 34 GPT-5.6 Luna max 72.1%
- 35 GPT-5.6 Terra max 71.7%
- 36
Gemini 3.1 Pro high 68.7%
- 37 Grok 4.20 reasoning 66.9%
- 38
Gemini 3 Flash high 66%
- 39 Beam 1.0 64.6%
- 40
Gemma 4 31B IT 61.6%
- 41 GPT-Rosalind-5.5 xhigh 61.4%
- 42 gpt-oss-120b 45.4%
- 43 Claude Haiku 4.5 16k 42%
- 44 gpt-oss-20b 39.3%
- 45
Qwen3.5 397B-A17B 35.7%
- 46 GLM-5 33.9%
- 47
Qwen3.7-Max 32.8%
- 48 GLM-5.2 max 32.6%
- 49 Kimi K3.1 Pro max 32.6%
- 50 GLM-5.3 max 32.5%
- 51
Mistral Large 4 high 32.2%
- 52
Nemotron 3.5 Super VL 32.1%
- 53
Qwen3.8-2.4T-A95B xhigh 31.6%
- 54
Inkling xhigh 30%
- 55 Kimi K3 max 27.3%
- 56
DeepSeek V4.1 Flash 1017 max 25.9%
- 57
Inkling-Small xhigh 24.9%
- 58 GLM-5.1 23.7%
- 59 GPT-4o Nov 2024 22.1%
- 60 Kimi K2.6 21.9%
- 61
Qwen 3.8 27B xhigh 21.9%
- 62
Llama 3.3 70B 21.2%
- 63 Kimi K2.5 21.2%
- 64
Nemotron 3.5 Lightning 30B-A3B 21.1%
- 65
Qwen3-235B-A22B-Thinking Jul 2025 20.9%
- 66
Nemotron 3 Nano 30B-A3B 20.2%
- 67
DeepSeek V4 Pro 0813 max 17.6%
- 68 GPT-4.1 16.7%
- 69 Kimi K2 Thinking 16.3%
- 70
DeepSeek-R1 May 2025 15.6%
- 71 GPT-4o Aug 2024 14.5%
- 72 GPT-4.1 mini 10.4%
- 73
Mistral Medium 3.1 10.3%
- 74
DeepSeek-V3.2 10.2%
- 75
Mistral Large 3 9.20%
Sources: Official BioTIER Refuse leaderboard and labs’ model cards
| # | Entry | Score ± | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Claude Opus 5.5 | 99.1% ±0.2 | max | Sep 2026 | Official leaderboard |
| 2 | Claude Sonnet 4.5 | 98.2% 98 to 98.3 | 16k | Sep 2025 | Official leaderboard |
| 3 | Claude Haiku 5.5 | 97.9% 97.6 to 98.1 | max | Oct 2026 | Official leaderboard |
| 4 | Claude Sonnet 5 | 96.1% ±0.2 | max | Jun 2026 | Official leaderboard |
| 5 | Claude Opus 4.8 | 95.9% 95.6 to 96.1 | max | May 2026 | Official leaderboard |
| 6 | Claude Sonnet 5.5 | 95.7% ±0.4 | max | Sep 2026 | Official leaderboard |
| 7 | | 95.5% 95.3 to 95.6 | xhigh | Oct 2026 | Official leaderboard |
| 8 | Claude Sonnet 4.6 | 95.2% ±0.2 | max | Feb 2026 | Official leaderboard |
| 9 | Claude Opus 4.7 | 94.9% 94.6 to 95.1 | max | Apr 2026 | Official leaderboard |
| 10 | Claude Opus 5 | 94.9% ±0.1 | max | Jul 2026 | Official leaderboard |
| 11 | Claude Opus 4.6 | 94.4% 94.3 to 94.6 | max | Feb 2026 | Official leaderboard |
| 12 | Claude Opus 4.5 | 92.6% ±0.2 | 16k | Nov 2025 | Official leaderboard |
| 13 | | 87.8% 87.6 to 88.1 | high | Sep 2026 | Official leaderboard |
| 14 | | 81.7% 81.3 to 82 | high | May 2026 | Official leaderboard |
| 15 | | 81.1% ±0.3 | high | Jul 2026 | Official leaderboard |
| 16 | Grok 4.7 | 80.3% ±0.6 | xhigh | Sep 2026 | Official leaderboard |
| 17 | GPT-6 Sol | 79.8% 79.2 to 80.3 | max | Sep 2026 | Official leaderboard |
| 18 | GPT-6 Luna | 79.5% 78.7 to 80.2 | max | Sep 2026 | Official leaderboard |
| 19 | GPT-6.1 Terra | 79.1% 78.3 to 79.8 | max | Oct 2026 | Official leaderboard |
| 20 | | 78.2% 77.7 to 78.8 | high | Aug 2026 | Official leaderboard |
| 21 | GPT-6 Astra | 78.1% ±0.4 | max | Sep 2026 | Official leaderboard |
| 22 | | 78% ±0.6 | high | Sep 2026 | Official leaderboard |
| 23 | GPT-6.1 Sol | 77.8% ±0.7 | max | Sep 2026 | Official leaderboard |
| 24 | GPT-5.4 | 76.9% ±0.7 | xhigh | Mar 2026 | Official leaderboard |
| 25 | | 76.3% 75.8 to 76.9 | xhigh | Aug 2026 | Official leaderboard |
| 26 | Grok 4.5 | 76.2% ±0.4 | high | Jul 2026 | Official leaderboard |
| 27 | Grok 4.6 | 76.1% ±0.4 | xhigh | Aug 2026 | Official leaderboard |
| 28 | GPT-5.2 | 75.4% ±0.6 | xhigh | Dec 2025 | Official leaderboard |
| 29 | GPT-5.5 | 75.3% ±0.5 | xhigh | Apr 2026 | Official leaderboard |
| 30 | GPT-5.6 Sol | 74% 73.2 to 74.9 | max | Jul 2026 | Official leaderboard |
| 31 | GPT-5.5 Pro | 73.6% 72.9 to 74.4 | xhigh | Apr 2026 | Official leaderboard |
| 32 | GPT-5.6 Luna | 72.1% ±0.8 | max | Jul 2026 | Official leaderboard |
| 33 | GPT-5.6 Terra | 71.7% 71 to 72.5 | max | Jul 2026 | Official leaderboard |
| 34 | | 68.7% 68.3 to 69 | high | Feb 2026 | Official leaderboard |
| 35 | Grok 4.20 | 66.9% 66.1 to 67.8 | reasoning | Feb 2026 | Official leaderboard |
| 36 | | 66% 65.3 to 66.6 | high | Dec 2025 | Official leaderboard |
| 37 | | 61.6% 61.2 to 61.9 | Apr 2026 | Official leaderboard | |
| 38 | GPT-Rosalind-5.5 | 61.4% 60.7 to 62 | xhigh | Jun 2026 | Official leaderboard |
| 39 | Beam 1.0 | 61.3% ±1 | max | Oct 2026 | Official leaderboard |
| 40 | gpt-oss-120b | 45.4% 44.9 to 46 | Aug 2025 | Official leaderboard | |
| 41 | Claude Haiku 4.5 | 42% ±0.6 | 16k | Oct 2025 | Official leaderboard |
| 42 | gpt-oss-20b | 39.3% ±2.9 | Aug 2025 | Official leaderboard | |
| 43 | | 35.7% ±0.4 | Feb 2026 | Official leaderboard | |
| 44 | GLM-5 | 33.9% 32.8 to 34.9 | Feb 2026 | Official leaderboard | |
| 45 | | 32.8% 32.1 to 33.4 | May 2026 | Official leaderboard | |
| 46 | GLM-5.2 | 32.6% 31.8 to 33.3 | max | Jun 2026 | Official leaderboard |
| 47 | Kimi K3.1 Pro | 32.6% 32 to 33.3 | max | Oct 2026 | Official leaderboard |
| 48 | GLM-5.3 | 32.5% 31.9 to 33.2 | max | Aug 2026 | Official leaderboard |
| 49 | Nemotron 3 Ultra 550B A55B open weights, 550B | 32.4% ±1 | Official leaderboard | ||
| 50 | | 32.2% ±0.7 | high | Oct 2026 | Official leaderboard |
| 51 | | 32.1% 31.2 to 33.1 | Oct 2026 | Official leaderboard | |
| 52 | | 31.6% ±0.6 | xhigh | Aug 2026 | Official leaderboard |
| 53 | | 30% ±0.9 | xhigh | Jul 2026 | Official leaderboard |
| 54 | Kimi K3 | 27.3% ±0.8 | max | Jul 2026 | Official leaderboard |
| 55 | | 25.9% 25.1 to 26.6 | max | Oct 2026 | Official leaderboard |
| 56 | | 24.9% 24.3 to 25.6 | xhigh | Jul 2026 | Official leaderboard |
| 57 | Nemotron 3 Super 120B A12B open weights, 120B | 23.9% ±0.9 | Official leaderboard | ||
| 58 | GLM-5.1 | 23.7% ±0.7 | Apr 2026 | Official leaderboard | |
| 59 | GPT-4o Nov 2024 | 22.1% 21.7 to 22.4 | Nov 2024 | Official leaderboard | |
| 60 | Kimi K2.6 | 21.9% ±0.5 | Apr 2026 | Official leaderboard | |
| 61 | | 21.9% ±0.8 | xhigh | Aug 2026 | Official leaderboard |
| 62 | | 21.2% 20.9 to 21.6 | Dec 2024 | Official leaderboard | |
| 63 | Kimi K2.5 | 21.2% 20.8 to 21.5 | Jan 2026 | Official leaderboard | |
| 64 | | 21.1% ±0.5 | Aug 2026 | Official leaderboard | |
| 65 | | 20.9% 20.1 to 21.6 | Jul 2025 | Official leaderboard | |
| 66 | | 20.2% ±0.6 | Dec 2025 | Official leaderboard | |
| 67 | | 17.6% ±0.5 | max | Aug 2026 | Official leaderboard |
| 68 | GPT-4.1 | 16.7% 16.4 to 17.1 | Apr 2025 | Official leaderboard | |
| 69 | Kimi K2 Thinking | 16.3% ±0.4 | Nov 2025 | Official leaderboard | |
| 70 | | 15.6% ±13 | May 2025 | Official leaderboard | |
| 71 | GPT-4o mini (snapshot unspecified) closed weights | 15.4% 14.8 to 15.9 | Official leaderboard | ||
| 72 | GPT-4o Aug 2024 | 14.5% ±0.7 | Aug 2024 | Official leaderboard | |
| 73 | GPT-4.1 mini | 10.4% ±0.4 | Apr 2025 | Official leaderboard | |
| 74 | | 10.3% 10 to 10.5 | Aug 2025 | Official leaderboard | |
| 75 | | 10.2% 9.6 to 10.7 | Dec 2025 | Official leaderboard | |
| 76 | Gemini 2.5 Flash (snapshot unspecified) closed weights | 9.30% 8.7 to 10 | 16k | Official leaderboard | |
| 77 | | 9.20% ±0.3 | Dec 2025 | Official leaderboard |
| # | Model | Score ± | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Claude Opus 5.5 | 99.1% ±0.2 | max | Sep 2026 | Official leaderboard |
| 2 | Claude Sonnet 4.5 | 98.2% 98 to 98.3 | 16k | Sep 2025 | Official leaderboard |
| 3 | Claude Haiku 5.5 | 97.9% 97.6 to 98.1 | max | Oct 2026 | Official leaderboard |
| 4 | Claude Sonnet 5 | 96.1% ±0.2 | max | Jun 2026 | Official leaderboard |
| 5 | Claude Opus 4.8 | 95.9% 95.6 to 96.1 | max | May 2026 | Official leaderboard |
| 6 | Claude Sonnet 5.5 | 95.7% ±0.4 | max | Sep 2026 | Official leaderboard |
| 7 | | 95.5% 95.3 to 95.6 | xhigh | Oct 2026 | Official leaderboard |
| 8 | Claude Sonnet 4.6 | 95.2% ±0.2 | max | Feb 2026 | Official leaderboard |
| 9 | Claude Opus 4.7 | 94.9% 94.6 to 95.1 | max | Apr 2026 | Official leaderboard |
| 10 | Claude Opus 5 | 94.9% ±0.1 | max | Jul 2026 | Official leaderboard |
| 11 | Claude Opus 4.6 | 94.4% 94.3 to 94.6 | max | Feb 2026 | Official leaderboard |
| 12 | Claude Opus 4.5 | 92.6% ±0.2 | 16k | Nov 2025 | Official leaderboard |
| 13 | | 87.8% 87.6 to 88.1 | high | Sep 2026 | Official leaderboard |
| 14 | | 81.7% 81.3 to 82 | high | May 2026 | Official leaderboard |
| 15 | | 81.1% ±0.3 | high | Jul 2026 | Official leaderboard |
| 16 | Grok 4.7 | 80.3% ±0.6 | xhigh | Sep 2026 | Official leaderboard |
| 17 | GPT-6 Sol | 79.8% 79.2 to 80.3 | max | Sep 2026 | Official leaderboard |
| 18 | GPT-6 Luna | 79.5% 78.7 to 80.2 | max | Sep 2026 | Official leaderboard |
| 19 | GPT-6.1 Terra | 79.1% 78.3 to 79.8 | max | Oct 2026 | Official leaderboard |
| 20 | | 78.2% 77.7 to 78.8 | high | Aug 2026 | Official leaderboard |
| 21 | GPT-6 Astra | 78.1% ±0.4 | max | Sep 2026 | Official leaderboard |
| 22 | | 78% ±0.6 | high | Sep 2026 | Official leaderboard |
| 23 | GPT-6.1 Sol | 77.8% ±0.7 | max | Sep 2026 | Official leaderboard |
| 24 | GPT-5.4 | 76.9% ±0.7 | xhigh | Mar 2026 | Official leaderboard |
| 25 | | 76.3% 75.8 to 76.9 | xhigh | Aug 2026 | Official leaderboard |
| 26 | Grok 4.5 | 76.2% ±0.4 | high | Jul 2026 | Official leaderboard |
| 27 | Grok 4.6 | 76.1% ±0.4 | xhigh | Aug 2026 | Official leaderboard |
| 28 | GPT-5.2 | 75.4% ±0.6 | xhigh | Dec 2025 | Official leaderboard |
| 29 | GPT-5.5 | 75.3% ±0.5 | xhigh | Apr 2026 | Official leaderboard |
| 30 | GPT-5.6 Sol | 74% 73.2 to 74.9 | max | Jul 2026 | Official leaderboard |
| 31 | GPT-5.5 Pro | 73.6% 72.9 to 74.4 | xhigh | Apr 2026 | Official leaderboard |
| 32 | GPT-5.6 Luna | 72.1% ±0.8 | max | Jul 2026 | Official leaderboard |
| 33 | GPT-5.6 Terra | 71.7% 71 to 72.5 | max | Jul 2026 | Official leaderboard |
| 34 | | 68.7% 68.3 to 69 | high | Feb 2026 | Official leaderboard |
| 35 | Grok 4.20 | 66.9% 66.1 to 67.8 | reasoning | Feb 2026 | Official leaderboard |
| 36 | | 66% 65.3 to 66.6 | high | Dec 2025 | Official leaderboard |
| 37 | | 61.6% 61.2 to 61.9 | Apr 2026 | Official leaderboard | |
| 38 | GPT-Rosalind-5.5 | 61.4% 60.7 to 62 | xhigh | Jun 2026 | Official leaderboard |
| 39 | Beam 1.0 | 61.3% ±1 | max | Oct 2026 | Official leaderboard |
| 40 | gpt-oss-120b | 45.4% 44.9 to 46 | Aug 2025 | Official leaderboard | |
| 41 | Claude Haiku 4.5 | 42% ±0.6 | 16k | Oct 2025 | Official leaderboard |
| 42 | gpt-oss-20b | 39.3% ±2.9 | Aug 2025 | Official leaderboard | |
| 43 | | 35.7% ±0.4 | Feb 2026 | Official leaderboard | |
| 44 | GLM-5 | 33.9% 32.8 to 34.9 | Feb 2026 | Official leaderboard | |
| 45 | | 32.8% 32.1 to 33.4 | May 2026 | Official leaderboard | |
| 46 | GLM-5.2 | 32.6% 31.8 to 33.3 | max | Jun 2026 | Official leaderboard |
| 47 | Kimi K3.1 Pro | 32.6% 32 to 33.3 | max | Oct 2026 | Official leaderboard |
| 48 | GLM-5.3 | 32.5% 31.9 to 33.2 | max | Aug 2026 | Official leaderboard |
| 49 | | 32.2% ±0.7 | high | Oct 2026 | Official leaderboard |
| 50 | | 32.1% 31.2 to 33.1 | Oct 2026 | Official leaderboard | |
| 51 | | 31.6% ±0.6 | xhigh | Aug 2026 | Official leaderboard |
| 52 | | 30% ±0.9 | xhigh | Jul 2026 | Official leaderboard |
| 53 | Kimi K3 | 27.3% ±0.8 | max | Jul 2026 | Official leaderboard |
| 54 | | 25.9% 25.1 to 26.6 | max | Oct 2026 | Official leaderboard |
| 55 | | 24.9% 24.3 to 25.6 | xhigh | Jul 2026 | Official leaderboard |
| 56 | GLM-5.1 | 23.7% ±0.7 | Apr 2026 | Official leaderboard | |
| 57 | GPT-4o Nov 2024 | 22.1% 21.7 to 22.4 | Nov 2024 | Official leaderboard | |
| 58 | Kimi K2.6 | 21.9% ±0.5 | Apr 2026 | Official leaderboard | |
| 59 | | 21.9% ±0.8 | xhigh | Aug 2026 | Official leaderboard |
| 60 | | 21.2% 20.9 to 21.6 | Dec 2024 | Official leaderboard | |
| 61 | Kimi K2.5 | 21.2% 20.8 to 21.5 | Jan 2026 | Official leaderboard | |
| 62 | | 21.1% ±0.5 | Aug 2026 | Official leaderboard | |
| 63 | | 20.9% 20.1 to 21.6 | Jul 2025 | Official leaderboard | |
| 64 | | 20.2% ±0.6 | Dec 2025 | Official leaderboard | |
| 65 | | 17.6% ±0.5 | max | Aug 2026 | Official leaderboard |
| 66 | GPT-4.1 | 16.7% 16.4 to 17.1 | Apr 2025 | Official leaderboard | |
| 67 | Kimi K2 Thinking | 16.3% ±0.4 | Nov 2025 | Official leaderboard | |
| 68 | | 15.6% ±13 | May 2025 | Official leaderboard | |
| 69 | GPT-4o Aug 2024 | 14.5% ±0.7 | Aug 2024 | Official leaderboard | |
| 70 | GPT-4.1 mini | 10.4% ±0.4 | Apr 2025 | Official leaderboard | |
| 71 | | 10.3% 10 to 10.5 | Aug 2025 | Official leaderboard | |
| 72 | | 10.2% 9.6 to 10.7 | Dec 2025 | Official leaderboard | |
| 73 | | 9.20% ±0.3 | Dec 2025 | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Claude Opus 5.5 | 99.1% | max | Sep 2026 | Official leaderboard |
| 2 | 98.7% | Oct 2026 | Lab model card | ||
| 3 | Claude Sonnet 4.5 | 98.2% | 16k | Sep 2025 | Official leaderboard |
| 4 | Claude Haiku 5.5 | 97.9% | max | Oct 2026 | Official leaderboard |
| 5 | Claude Sonnet 5 | 96.1% | max | Jun 2026 | Official leaderboard |
| 6 | Claude Opus 4.8 | 95.9% | max | May 2026 | Official leaderboard |
| 7 | Claude Sonnet 5.5 | 95.7% | max | Sep 2026 | Official leaderboard |
| 8 | 95.5% | xhigh | Oct 2026 | Official leaderboard | |
| 9 | Claude Sonnet 4.6 | 95.2% | max | Feb 2026 | Official leaderboard |
| 10 | Claude Opus 4.7 | 94.9% | max | Apr 2026 | Official leaderboard |
| 11 | Claude Opus 5 | 94.9% | max | Jul 2026 | Official leaderboard |
| 12 | Claude Opus 4.6 | 94.4% | max | Feb 2026 | Official leaderboard |
| 13 | Claude Opus 4.5 | 92.6% | 16k | Nov 2025 | Official leaderboard |
| 14 | 87.8% | high | Sep 2026 | Official leaderboard | |
| 15 | GPT-6 Astra | 82.9% | Sep 2026 | Lab model card | |
| 16 | 81.7% | high | May 2026 | Official leaderboard | |
| 17 | 81.1% | high | Jul 2026 | Official leaderboard | |
| 18 | Grok 4.7 | 80.3% | xhigh | Sep 2026 | Official leaderboard |
| 19 | GPT-6.2 Astra | 79.8% | Nov 2026 | Lab model card | |
| 20 | GPT-6 Sol | 79.8% | max | Sep 2026 | Official leaderboard |
| 21 | GPT-6 Luna | 79.5% | max | Sep 2026 | Official leaderboard |
| 22 | GPT-6.1 Terra | 79.1% | max | Oct 2026 | Official leaderboard |
| 23 | 78.2% | high | Aug 2026 | Official leaderboard | |
| 24 | 78% | high | Sep 2026 | Official leaderboard | |
| 25 | GPT-6.1 Sol | 77.8% | max | Sep 2026 | Official leaderboard |
| 26 | GPT-5.4 | 76.9% | xhigh | Mar 2026 | Official leaderboard |
| 27 | 76.3% | xhigh | Aug 2026 | Official leaderboard | |
| 28 | Grok 4.5 | 76.2% | high | Jul 2026 | Official leaderboard |
| 29 | Grok 4.6 | 76.1% | xhigh | Aug 2026 | Official leaderboard |
| 30 | GPT-5.2 | 75.4% | xhigh | Dec 2025 | Official leaderboard |
| 31 | GPT-5.5 | 75.3% | xhigh | Apr 2026 | Official leaderboard |
| 32 | GPT-5.6 Sol | 74% | max | Jul 2026 | Official leaderboard |
| 33 | GPT-5.5 Pro | 73.6% | xhigh | Apr 2026 | Official leaderboard |
| 34 | GPT-5.6 Luna | 72.1% | max | Jul 2026 | Official leaderboard |
| 35 | GPT-5.6 Terra | 71.7% | max | Jul 2026 | Official leaderboard |
| 36 | 68.7% | high | Feb 2026 | Official leaderboard | |
| 37 | Grok 4.20 | 66.9% | reasoning | Feb 2026 | Official leaderboard |
| 38 | 66% | high | Dec 2025 | Official leaderboard | |
| 39 | Beam 1.0 | 64.6% | Oct 2026 | Lab model card | |
| 40 | 61.6% | Apr 2026 | Official leaderboard | ||
| 41 | GPT-Rosalind-5.5 | 61.4% | xhigh | Jun 2026 | Official leaderboard |
| 42 | gpt-oss-120b | 45.4% | Aug 2025 | Official leaderboard | |
| 43 | Claude Haiku 4.5 | 42% | 16k | Oct 2025 | Official leaderboard |
| 44 | gpt-oss-20b | 39.3% | Aug 2025 | Official leaderboard | |
| 45 | 35.7% | Feb 2026 | Official leaderboard | ||
| 46 | GLM-5 | 33.9% | Feb 2026 | Official leaderboard | |
| 47 | 32.8% | May 2026 | Official leaderboard | ||
| 48 | GLM-5.2 | 32.6% | max | Jun 2026 | Official leaderboard |
| 49 | Kimi K3.1 Pro | 32.6% | max | Oct 2026 | Official leaderboard |
| 50 | GLM-5.3 | 32.5% | max | Aug 2026 | Official leaderboard |
| 51 | 32.2% | high | Oct 2026 | Official leaderboard | |
| 52 | 32.1% | Oct 2026 | Official leaderboard | ||
| 53 | 31.6% | xhigh | Aug 2026 | Official leaderboard | |
| 54 | 30% | xhigh | Jul 2026 | Official leaderboard | |
| 55 | Kimi K3 | 27.3% | max | Jul 2026 | Official leaderboard |
| 56 | 25.9% | max | Oct 2026 | Official leaderboard | |
| 57 | 24.9% | xhigh | Jul 2026 | Official leaderboard | |
| 58 | GLM-5.1 | 23.7% | Apr 2026 | Official leaderboard | |
| 59 | GPT-4o Nov 2024 | 22.1% | Nov 2024 | Official leaderboard | |
| 60 | Kimi K2.6 | 21.9% | Apr 2026 | Official leaderboard | |
| 61 | 21.9% | xhigh | Aug 2026 | Official leaderboard | |
| 62 | 21.2% | Dec 2024 | Official leaderboard | ||
| 63 | Kimi K2.5 | 21.2% | Jan 2026 | Official leaderboard | |
| 64 | 21.1% | Aug 2026 | Official leaderboard | ||
| 65 | 20.9% | Jul 2025 | Official leaderboard | ||
| 66 | 20.2% | Dec 2025 | Official leaderboard | ||
| 67 | 17.6% | max | Aug 2026 | Official leaderboard | |
| 68 | GPT-4.1 | 16.7% | Apr 2025 | Official leaderboard | |
| 69 | Kimi K2 Thinking | 16.3% | Nov 2025 | Official leaderboard | |
| 70 | 15.6% | May 2025 | Official leaderboard | ||
| 71 | GPT-4o Aug 2024 | 14.5% | Aug 2024 | Official leaderboard | |
| 72 | GPT-4.1 mini | 10.4% | Apr 2025 | Official leaderboard | |
| 73 | 10.3% | Aug 2025 | Official leaderboard | ||
| 74 | 10.2% | Dec 2025 | Official leaderboard | ||
| 75 | 9.20% | Dec 2025 | Official leaderboard |
Labs report scores for 4 models in their cards. Switch the scores above to rank them together.
2 models have held number one on the official leaderboard since its release in March 2026 2 models have held number one with claims included since its release in March 2026
Lab claims count from the day the card came out, independent runs from the model’s release. We draw one row each for the benchmark’s own leaderboard and what labs claim in their cards.
Sources: Official BioTIER Refuse leaderboard and labs’ model cards
Claude Sonnet 4.5 on the official leaderboard passed 85% in September 2025
Models released before it came out already scored that high. All the scores we have, one dot each, so a model tested at several settings appears more than once.
Sources: Official BioTIER Refuse leaderboard and labs’ model cards
Labs report it only in system cards
We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark. Some system cards do, below.
4 system cards report it, from 3 labs
Besides launch posts, we read the system cards, model cards and technical reports labs publish with a release. They report far more, including safety and frontier-risk tests.
| Date | Document | Score | As the document writes it | Section |
|---|---|---|---|---|
| Nov 17, 2026 | GPT-6.2 Astra system card | 79.8 | SecureBio BioTIER, refusal rate on BioTIER-refuse with bio classifiers enabled, updated checkpoint (50.18% model-level with classifiers disabled, BioTIER-permit compliance 98.96%) (2.2.6.2) | Safety |
| Oct 12, 2026 | | 98.7 | BioTIER English refusal rate (CBRN Safety Evaluations table, section 4). Multilingual 98.7, adversarial 98.2 | Bio and chem risk |
| Oct 5, 2026 | Beam 1.0 technical report | 64.6 | BioTIER refuse split accuracy | Bio and chem risk |
| Sep 3, 2026 | GPT-6 Astra system card | 82.9 | SecureBio BioTIER, safeguarded configuration refused requests designated for refusal (benign allowed 95.3%) (10.2.4.2) | Safety |
The lab's number comes out ahead in 2 of 2 pairs that differ
Each score a lab reported for its own model, beside the same model's best score on the leaderboard. Gaps mostly come from different harnesses, effort levels and task subsets. Every benchmark.
More in Safety and alignment
- FORTRESS In use Paired adversarial and benign national-security misuse prompts. Rubric-graded risk score on adversarial and over-refusal score on benign prompts. Claude Opus 5.5 5.78%
- StrongREJECT In use Forbidden-prompt jailbreak benchmark with a rubric grader that scores how useful a non-refusal actually is.
Inkling 98.6%
- Gray Swan IPI Arena benchmark In use Indirect prompt-injection attacks from public red-teaming competitions replayed on tool, coding and browser agents. Attack success rate at k attempts.
Gemini 4 Carbon 0%
- Covert misbehavior (Thinking Machines) In use Agentic tasks that check whether a model misbehaves and conceals it. Scored by the rate of covert misbehavior.
Inkling 2 Mini 8.30%
- Destructive actions (Thinking Machines) In use Agentic tasks that check whether a model takes destructive actions it was not authorized to take. Scored by the rate of destructive actions.
Inkling 2 Small 4.70%
- Misbehavior (Thinking Machines) In use Agentic tasks that check whether a model fabricates results, ignores instructions or hides what it did when a task is hard. Scored by the rate of misbehavior.
Inkling 2 Mini 9.80%
Data audit
All 11 checks pass. Every benchmark
Catalog record 2 pass
- Pass Task access The tasks are private.
- Pass Confidence Checked against a primary source.
Scores 4 pass 1 note
- Pass Independent runs Scores from the official leaderboard (77 entries) and system cards (4 scores).
- Pass Models scored 75 models we track, plus 4 other entries on its leaderboard.
- Pass Newest score The newest is from November 2026, for GPT-6.2 Astra.
- Pass Frontier models 7 of the top 10 have a score, led by GPT-6.2 Astra (#1).
- Note Headroom Saturated. The best is 99.1%, past 85% since September 2025.
Leaderboard import 3 pass
- Pass Importer We read the board on securebio.org: 73 rows for models we track and 4 other entries at the last read.
- Pass Last read November 18th, 2026.
- Pass Read worked The last read worked.
Consistency 2 pass
- Pass Lab claims Both pairs agree within 10 points.
- Pass Known issues No published audit found problems.