RiskRubric
Evaluators score public models on six dimensions of risk. The scorecard combines the results into a letter grade.
- Released
- August 2025
- Kind
- Deception and alignment
- Level
- Expert
- Scored
- Graded by people
- Answers
- Open-ended
- Tasks
- All public
- Measures
- Propensity
- Risk
- Privacy
- Behavior
- Harmful use, Lawlessness
- Language
- English
- Interaction
- Single turn
- Run by
- A host
- In views
- Content safety
- Strongest model tested
- GPT-6 Astra #4 on the capability index
- Last model added
- October 2026
- Scores refreshed
- November 20th, 2026
Claude Haiku 5.5 leads with 891 Claude Haiku 5.5 leads with 891
Best score per entry from the benchmark's own leaderboard, read on November 20th, 2026. Where a model ran with several settings or harnesses, we keep its best. Every run is in the next chart. 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.
- 891 Claude Haiku 5.5
- 872 Claude Opus 4.7
- 871 Claude Opus 5.5
- 869 Claude Haiku 4.5
- 868 Claude Opus 5
- 863 GPT-5.5
- 862 Claude Sonnet 5
- 859 GPT-5.4
- 856 GPT-6 Luna
- 847 Grok 4.6
- 827 Claude Sonnet 5.5
- 824 Kimi K3
- 824 Claude Opus 4.8
- 823 GPT-6.1 Sol
- 822 GPT-5.6 Luna
- 819 GPT-5.6 Terra
- 819 GPT-6 Astra
- 816 GPT-5.6 Sol
- 799 Claude Sonnet 4.6
- 783
MiniMax-M2.5
- 776
MiniMax-M2.1
- 773 Grok 4.7
- 772 GPT-6.1 Terra
- 726 Kimi K2.6
- 726
Nova 2 Lite
- 701
Gemini 3.1 Pro
- 701
Gemini 3.5 Flash
- 690 Kimi K2.5
- 649 NVIDIA Nemotron Super 3
- 635 GLM-5
- 630
DeepSeek-V3.2
- 620 DeekSeep v3.1
- 618
Qwen3-Coder-Next
- 606
Nemotron 3 Nano 30B-A3B
- 597 GLM-4.7
- 591
Pixtral Large
- 561 Devstral 2 123B
- 561
Ministral 3B
Bars start at 500.
- 1 Claude Haiku 5.5 AWS Bedrock / 20261007 / Grade B 891
- 2 Claude Opus 4.7 AWS Bedrock / 20260416 / Grade B 872
- 3 Claude Opus 5.5 AWS Bedrock / 20260922 / Grade B 871
- 4 Claude Haiku 4.5 AWS Bedrock / 20251015 / Grade B 869
- 5 Claude Opus 5 AWS Bedrock / 20260724 / Grade B 868
- 6 GPT-5.5 AWS Bedrock / 20260922 / Grade B 863
- 7 Claude Sonnet 5 AWS Bedrock / 20260630 / Grade B 862
- 8 GPT-5.4 AWS Bedrock / 20260922 / Grade B 859
- 9 GPT-6 Luna AWS Bedrock / 20260922 / Grade B 856
- 10 Grok 4.6 AWS Bedrock / 20260819 / Grade B 847
- 11 Claude Sonnet 5.5 AWS Bedrock / 20260928 / Grade B 827
- 12 Kimi K3 AWS Bedrock / 20260827 / Grade B 824
- 13 Claude Opus 4.8 AWS Bedrock / 20260528 / Grade B 824
- 14 GPT-6.1 Sol AWS Bedrock / 20261015 / Grade B 823
- 15 GPT-5.6 Luna AWS Bedrock / 20260813 / Grade B 822
- 16 GPT-5.6 Terra AWS Bedrock / 20260813 / Grade B 819
- 17 GPT-6 Astra AWS Bedrock / 20260908 / Grade B 819
- 18 GPT-5.6 Sol AWS Bedrock / 20260813 / Grade B 816
- 19 Claude Sonnet 4.6 Direct API / 20250514 / Grade C 799
- 20
MiniMax-M2.5 AWS Bedrock / 20260318 / Grade C 783
- 21
MiniMax-M2.1 AWS Bedrock / 20250602 / Grade C 776
- 22 Grok 4.7 AWS Bedrock / 20260928 / Grade C 773
- 23 GPT-6.1 Terra AWS Bedrock / 20261026 / Grade C 772
- 24 Kimi K2.6 Moonshot AI / 2.6 / Grade C 726
- 25
Nova 2 Lite AWS Bedrock / 20251202 / Grade C 726
- 26
Gemini 3.1 Pro Google Cloud Vertex AI / 3.1 / Grade C 701
- 27
Gemini 3.5 Flash Google Cloud Vertex AI / 3.5 / Grade C 701
- 28 Kimi K2.5 AWS Bedrock / 20260206 / Grade D 690
- 29 NVIDIA Nemotron Super 3 AWS Bedrock / 20260318 / Grade D 649
- 30 GLM-5 AWS Bedrock / 20260318 / Grade D 635
- 31
DeepSeek-V3.2 AWS Bedrock / 20260206 / Grade D 630
- 32 DeekSeep v3.1 AWS Bedrock / 20250918 / Grade D 620
- 33
Qwen3-Coder-Next AWS Bedrock / 20260206 / Grade D 618
- 34
Nemotron 3 Nano 30B-A3B AWS Bedrock / 20251223 / Grade D 606
- 35 GLM-4.7 AWS Bedrock / 20260206 / Grade F 597
- 36
Pixtral Large AWS Bedrock / 20250408 / Grade F 591
- 37 Devstral 2 123B AWS Bedrock / 20260217 / Grade F 561
- 38
Ministral 3B AWS Bedrock / 20251202 / Grade F 561
Source: Official RiskRubric leaderboard
- 891 Claude Haiku 5.5
- 872 Claude Opus 4.7
- 871 Claude Opus 5.5
- 869 Claude Haiku 4.5
- 868 Claude Opus 5
- 863 GPT-5.5
- 862 Claude Sonnet 5
- 859 GPT-5.4
- 856 GPT-6 Luna
- 847 Grok 4.6
- 827 Claude Sonnet 5.5
- 824 Kimi K3
- 824 Claude Opus 4.8
- 823 GPT-6.1 Sol
- 822 GPT-5.6 Luna
- 819 GPT-5.6 Terra
- 819 GPT-6 Astra
- 816 GPT-5.6 Sol
- 799 Claude Sonnet 4.6
- 783
MiniMax-M2.5
- 776
MiniMax-M2.1
- 773 Grok 4.7
- 772 GPT-6.1 Terra
- 726 Kimi K2.6
- 726
Nova 2 Lite
- 701
Gemini 3.1 Pro
- 701
Gemini 3.5 Flash
- 690 Kimi K2.5
- 635 GLM-5
- 630
DeepSeek-V3.2
- 618
Qwen3-Coder-Next
- 606
Nemotron 3 Nano 30B-A3B
- 597 GLM-4.7
- 591
Pixtral Large
- 561
Ministral 3B
Bars start at 500.
- 1 Claude Haiku 5.5 AWS Bedrock / 20261007 / Grade B 891
- 2 Claude Opus 4.7 AWS Bedrock / 20260416 / Grade B 872
- 3 Claude Opus 5.5 AWS Bedrock / 20260922 / Grade B 871
- 4 Claude Haiku 4.5 AWS Bedrock / 20251015 / Grade B 869
- 5 Claude Opus 5 AWS Bedrock / 20260724 / Grade B 868
- 6 GPT-5.5 AWS Bedrock / 20260922 / Grade B 863
- 7 Claude Sonnet 5 AWS Bedrock / 20260630 / Grade B 862
- 8 GPT-5.4 AWS Bedrock / 20260922 / Grade B 859
- 9 GPT-6 Luna AWS Bedrock / 20260922 / Grade B 856
- 10 Grok 4.6 AWS Bedrock / 20260819 / Grade B 847
- 11 Claude Sonnet 5.5 AWS Bedrock / 20260928 / Grade B 827
- 12 Kimi K3 AWS Bedrock / 20260827 / Grade B 824
- 13 Claude Opus 4.8 AWS Bedrock / 20260528 / Grade B 824
- 14 GPT-6.1 Sol AWS Bedrock / 20261015 / Grade B 823
- 15 GPT-5.6 Luna AWS Bedrock / 20260813 / Grade B 822
- 16 GPT-5.6 Terra AWS Bedrock / 20260813 / Grade B 819
- 17 GPT-6 Astra AWS Bedrock / 20260908 / Grade B 819
- 18 GPT-5.6 Sol AWS Bedrock / 20260813 / Grade B 816
- 19 Claude Sonnet 4.6 Direct API / 20250514 / Grade C 799
- 20
MiniMax-M2.5 AWS Bedrock / 20260318 / Grade C 783
- 21
MiniMax-M2.1 AWS Bedrock / 20250602 / Grade C 776
- 22 Grok 4.7 AWS Bedrock / 20260928 / Grade C 773
- 23 GPT-6.1 Terra AWS Bedrock / 20261026 / Grade C 772
- 24 Kimi K2.6 Moonshot AI / 2.6 / Grade C 726
- 25
Nova 2 Lite AWS Bedrock / 20251202 / Grade C 726
- 26
Gemini 3.1 Pro Google Cloud Vertex AI / 3.1 / Grade C 701
- 27
Gemini 3.5 Flash Google Cloud Vertex AI / 3.5 / Grade C 701
- 28 Kimi K2.5 AWS Bedrock / 20260206 / Grade D 690
- 29 GLM-5 AWS Bedrock / 20260318 / Grade D 635
- 30
DeepSeek-V3.2 AWS Bedrock / 20260206 / Grade D 630
- 31
Qwen3-Coder-Next AWS Bedrock / 20260206 / Grade D 618
- 32
Nemotron 3 Nano 30B-A3B AWS Bedrock / 20251223 / Grade D 606
- 33 GLM-4.7 AWS Bedrock / 20260206 / Grade F 597
- 34
Pixtral Large AWS Bedrock / 20250408 / Grade F 591
- 35
Ministral 3B AWS Bedrock / 20251202 / Grade F 561
Source: Official RiskRubric leaderboard
| # | Entry | Score | Setting | Released | Run on | Link | Source |
|---|---|---|---|---|---|---|---|
| 1 | Claude Haiku 5.5 | 891 | AWS Bedrock / 20261007 / Grade B | Oct 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 2 | Claude Opus 4.7 | 872 | AWS Bedrock / 20260416 / Grade B | Apr 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 3 | Claude Opus 5.5 | 871 | AWS Bedrock / 20260922 / Grade B | Sep 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 4 | Claude Haiku 4.5 | 869 | AWS Bedrock / 20251015 / Grade B | Oct 2025 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 5 | Claude Opus 5 | 868 | AWS Bedrock / 20260724 / Grade B | Jul 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 6 | GPT-5.5 | 863 | AWS Bedrock / 20260922 / Grade B | Apr 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 7 | Claude Sonnet 5 | 862 | AWS Bedrock / 20260630 / Grade B | Jun 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 8 | GPT-5.4 | 859 | AWS Bedrock / 20260922 / Grade B | Mar 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 9 | GPT-6 Luna | 856 | AWS Bedrock / 20260922 / Grade B | Sep 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 10 | Grok 4.6 | 847 | AWS Bedrock / 20260819 / Grade B | Aug 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 11 | Claude Sonnet 5.5 | 827 | AWS Bedrock / 20260928 / Grade B | Sep 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 12 | Kimi K3 | 824 | AWS Bedrock / 20260827 / Grade B | Jul 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 13 | Claude Opus 4.8 | 824 | AWS Bedrock / 20260528 / Grade B | May 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 14 | GPT-6.1 Sol | 823 | AWS Bedrock / 20261015 / Grade B | Sep 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 15 | GPT-5.6 Luna | 822 | AWS Bedrock / 20260813 / Grade B | Jul 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 16 | GPT-5.6 Terra | 819 | AWS Bedrock / 20260813 / Grade B | Jul 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 17 | GPT-6 Astra | 819 | AWS Bedrock / 20260908 / Grade B | Sep 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 18 | GPT-5.6 Sol | 816 | AWS Bedrock / 20260813 / Grade B | Jul 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 19 | Claude Sonnet 4.6 | 799 | Direct API / 20250514 / Grade C | Feb 2026 | Aug 2026 | riskrubric.ai | Official leaderboard |
| 20 | | 783 | AWS Bedrock / 20260318 / Grade C | Feb 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 21 | | 776 | AWS Bedrock / 20250602 / Grade C | Dec 2025 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 22 | Grok 4.7 | 773 | AWS Bedrock / 20260928 / Grade C | Sep 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 23 | GPT-6.1 Terra | 772 | AWS Bedrock / 20261026 / Grade C | Oct 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 24 | Kimi K2.6 | 726 | Moonshot AI / 2.6 / Grade C | Apr 2026 | Jul 2026 | riskrubric.ai | Official leaderboard |
| 25 | | 726 | AWS Bedrock / 20251202 / Grade C | Dec 2025 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 26 | | 701 | Google Cloud Vertex AI / 3.1 / Grade C | Feb 2026 | Jul 2026 | riskrubric.ai | Official leaderboard |
| 27 | | 701 | Google Cloud Vertex AI / 3.5 / Grade C | May 2026 | Jul 2026 | riskrubric.ai | Official leaderboard |
| 28 | Kimi K2.5 | 690 | AWS Bedrock / 20260206 / Grade D | Jan 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 29 | NVIDIA Nemotron Super 3 | 649 | AWS Bedrock / 20260318 / Grade D | Oct 2026 | riskrubric.ai | Official leaderboard | |
| 30 | GLM-5 | 635 | AWS Bedrock / 20260318 / Grade D | Feb 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 31 | | 630 | AWS Bedrock / 20260206 / Grade D | Dec 2025 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 32 | DeekSeep v3.1 | 620 | AWS Bedrock / 20250918 / Grade D | Oct 2026 | riskrubric.ai | Official leaderboard | |
| 33 | | 618 | AWS Bedrock / 20260206 / Grade D | Feb 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 34 | | 606 | AWS Bedrock / 20251223 / Grade D | Dec 2025 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 35 | GLM-4.7 | 597 | AWS Bedrock / 20260206 / Grade F | Dec 2025 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 36 | | 591 | AWS Bedrock / 20250408 / Grade F | Nov 2024 | Nov 2026 | riskrubric.ai | Official leaderboard |
| 37 | Devstral 2 123B | 561 | AWS Bedrock / 20260217 / Grade F | Oct 2026 | riskrubric.ai | Official leaderboard | |
| 38 | | 561 | AWS Bedrock / 20251202 / Grade F | Oct 2024 | Oct 2026 | riskrubric.ai | Official leaderboard |
| # | Model | Score | Setting | Released | Run on | Link | Source |
|---|---|---|---|---|---|---|---|
| 1 | Claude Haiku 5.5 | 891 | AWS Bedrock / 20261007 / Grade B | Oct 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 2 | Claude Opus 4.7 | 872 | AWS Bedrock / 20260416 / Grade B | Apr 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 3 | Claude Opus 5.5 | 871 | AWS Bedrock / 20260922 / Grade B | Sep 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 4 | Claude Haiku 4.5 | 869 | AWS Bedrock / 20251015 / Grade B | Oct 2025 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 5 | Claude Opus 5 | 868 | AWS Bedrock / 20260724 / Grade B | Jul 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 6 | GPT-5.5 | 863 | AWS Bedrock / 20260922 / Grade B | Apr 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 7 | Claude Sonnet 5 | 862 | AWS Bedrock / 20260630 / Grade B | Jun 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 8 | GPT-5.4 | 859 | AWS Bedrock / 20260922 / Grade B | Mar 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 9 | GPT-6 Luna | 856 | AWS Bedrock / 20260922 / Grade B | Sep 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 10 | Grok 4.6 | 847 | AWS Bedrock / 20260819 / Grade B | Aug 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 11 | Claude Sonnet 5.5 | 827 | AWS Bedrock / 20260928 / Grade B | Sep 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 12 | Kimi K3 | 824 | AWS Bedrock / 20260827 / Grade B | Jul 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 13 | Claude Opus 4.8 | 824 | AWS Bedrock / 20260528 / Grade B | May 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 14 | GPT-6.1 Sol | 823 | AWS Bedrock / 20261015 / Grade B | Sep 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 15 | GPT-5.6 Luna | 822 | AWS Bedrock / 20260813 / Grade B | Jul 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 16 | GPT-5.6 Terra | 819 | AWS Bedrock / 20260813 / Grade B | Jul 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 17 | GPT-6 Astra | 819 | AWS Bedrock / 20260908 / Grade B | Sep 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 18 | GPT-5.6 Sol | 816 | AWS Bedrock / 20260813 / Grade B | Jul 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 19 | Claude Sonnet 4.6 | 799 | Direct API / 20250514 / Grade C | Feb 2026 | Aug 2026 | riskrubric.ai | Official leaderboard |
| 20 | | 783 | AWS Bedrock / 20260318 / Grade C | Feb 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 21 | | 776 | AWS Bedrock / 20250602 / Grade C | Dec 2025 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 22 | Grok 4.7 | 773 | AWS Bedrock / 20260928 / Grade C | Sep 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 23 | GPT-6.1 Terra | 772 | AWS Bedrock / 20261026 / Grade C | Oct 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 24 | Kimi K2.6 | 726 | Moonshot AI / 2.6 / Grade C | Apr 2026 | Jul 2026 | riskrubric.ai | Official leaderboard |
| 25 | | 726 | AWS Bedrock / 20251202 / Grade C | Dec 2025 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 26 | | 701 | Google Cloud Vertex AI / 3.1 / Grade C | Feb 2026 | Jul 2026 | riskrubric.ai | Official leaderboard |
| 27 | | 701 | Google Cloud Vertex AI / 3.5 / Grade C | May 2026 | Jul 2026 | riskrubric.ai | Official leaderboard |
| 28 | Kimi K2.5 | 690 | AWS Bedrock / 20260206 / Grade D | Jan 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 29 | GLM-5 | 635 | AWS Bedrock / 20260318 / Grade D | Feb 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 30 | | 630 | AWS Bedrock / 20260206 / Grade D | Dec 2025 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 31 | | 618 | AWS Bedrock / 20260206 / Grade D | Feb 2026 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 32 | | 606 | AWS Bedrock / 20251223 / Grade D | Dec 2025 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 33 | GLM-4.7 | 597 | AWS Bedrock / 20260206 / Grade F | Dec 2025 | Oct 2026 | riskrubric.ai | Official leaderboard |
| 34 | | 591 | AWS Bedrock / 20250408 / Grade F | Nov 2024 | Nov 2026 | riskrubric.ai | Official leaderboard |
| 35 | | 561 | AWS Bedrock / 20251202 / Grade F | Oct 2024 | Oct 2026 | riskrubric.ai | Official leaderboard |
Rankings by part
We read 6 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as Transparency, Reliability and Security. Claude Sonnet 4.6 leads 1 of them, and 5 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.
Transparency
35 models we track have a score here.
- 1 Claude Sonnet 4.6 ↑18 820
- 2 Claude Sonnet 5 ↑5 748
- 3 Claude Opus 5 ↑2 701
- 4 Claude Opus 4.8 ↑9 695
- 5
Gemini 3.1 Pro ↑21 663
- 6 GPT-5.5 647
- 7 Kimi K2.6 ↑17 642
- 8 Claude Haiku 5.5 ↓7 616
- 9 Claude Haiku 4.5 ↓5 602
- 10 Claude Opus 4.7 ↓8 602
Reliability
35 models we track have a score here.
- 1 Claude Sonnet 5 ↑6 931
- 2 GPT-6.1 Sol ↑12 926
- 3 GPT-6.1 Terra ↑20 926
- 4 GPT-6 Astra ↑13 922
- 5 GPT-6 Luna ↑4 916
- 6 Claude Haiku 5.5 ↓5 913
- 7 GPT-5.6 Sol ↑11 909
- 8 Claude Opus 4.7 ↓6 905
- 9 Kimi K2.6 ↑15 901
- 10 GPT-5.5 ↓4 900
Security
35 models we track have a score here.
- 1 Claude Opus 5.5 ↑2 995
- 2 Claude Opus 5 ↑3 962
- 3 Claude Haiku 5.5 ↓2 952
- 4 GPT-5.4 ↑4 946
- 5 Claude Haiku 4.5 ↓1 943
- 6 GPT-5.5 929
- 7 Grok 4.6 ↑3 924
- 8 Claude Opus 4.7 ↓6 919
- 9 GPT-6 Luna 912
- 10 Claude Sonnet 5.5 ↑1 896
Privacy
35 models we track have a score here.
- 1 GPT-5.4 ↑7 1,000
- 2 GPT-5.5 ↑4 1,000
- 3 GPT-6 Luna ↑6 980
- 4 Claude Haiku 5.5 ↓3 980
- 5 Grok 4.6 ↑5 973
- 6 Claude Opus 5.5 ↓3 966
- 7 Claude Haiku 4.5 ↓3 959
- 8 Claude Opus 4.7 ↓6 946
- 9 GPT-5.6 Luna ↑6 946
- 10 Claude Sonnet 5 ↓3 939
Safety and societal risk
35 models we track have a score here.
- 1 Claude Haiku 5.5 933
- 2 Claude Haiku 4.5 ↑2 921
- 3 Claude Opus 4.7 ↓1 912
- 4 GPT-6.1 Sol ↑10 903
- 5 Claude Sonnet 5 ↑2 901
- 6 Claude Opus 5 ↓1 897
- 7 GPT-5.5 ↓1 894
- 8 GPT-6 Luna ↑1 894
- 9 GPT-5.4 ↓1 891
- 10 GPT-5.6 Terra ↑6 888
Excessive agency
35 models we track have a score here.
- 1
Gemini 3.1 Pro ↑25 1,000
- 2 GPT-5.4 ↑6 1,000
- 3 Kimi K2.6 ↑21 1,000
- 4 GPT-5.5 ↑2 1,000
- 5
Gemini 3.5 Flash ↑22 1,000
- 6 Claude Sonnet 5 ↑1 1,000
- 7 Claude Opus 5 ↓2 1,000
- 8 Claude Opus 5.5 ↓5 984
- 9 Claude Opus 4.8 ↑4 969
- 10 Claude Sonnet 4.6 ↑9 953
Scores at each effort level
Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.
Source: Official RiskRubric leaderboard
| Model | Effort | Harness | Score | Cost per task | Output tokens |
|---|---|---|---|---|---|
| Claude Haiku 5.5 | AWS Bedrock / 20261007 / Grade B | 891 | |||
| Claude Opus 4.7 | AWS Bedrock / 20260416 / Grade B | 872 | |||
| Claude Opus 5.5 | AWS Bedrock / 20260922 / Grade B | 871 | |||
| Claude Haiku 4.5 | AWS Bedrock / 20251015 / Grade B | 869 | |||
| Claude Opus 5 | AWS Bedrock / 20260724 / Grade B | 868 | |||
| GPT-5.5 | AWS Bedrock / 20260922 / Grade B | 863 | |||
| Claude Sonnet 5 | AWS Bedrock / 20260630 / Grade B | 862 | |||
| GPT-5.4 | AWS Bedrock / 20260922 / Grade B | 859 | |||
| GPT-6 Luna | AWS Bedrock / 20260922 / Grade B | 856 | |||
| Grok 4.6 | AWS Bedrock / 20260819 / Grade B | 847 | |||
| Claude Sonnet 5.5 | AWS Bedrock / 20260928 / Grade B | 827 | |||
| Kimi K3 | AWS Bedrock / 20260827 / Grade B | 824 | |||
| Claude Opus 4.8 | AWS Bedrock / 20260528 / Grade B | 824 | |||
| GPT-6.1 Sol | AWS Bedrock / 20261015 / Grade B | 823 | |||
| GPT-5.6 Luna | AWS Bedrock / 20260813 / Grade B | 822 | |||
| GPT-5.6 Terra | AWS Bedrock / 20260813 / Grade B | 819 | |||
| GPT-6 Astra | AWS Bedrock / 20260908 / Grade B | 819 | |||
| GPT-5.6 Sol | AWS Bedrock / 20260813 / Grade B | 816 | |||
| Claude Sonnet 4.6 | Direct API / 20250514 / Grade C | 799 | |||
| Claude Sonnet 4.6 | AWS Bedrock / 4.6 / Grade C | 792 | |||
| GPT-5.5 | OpenAI / 5.5 / Grade C | 789 | |||
| AWS Bedrock / 20260318 / Grade C | 783 | ||||
| GPT-5.4 | Direct API / 5.4 / Grade C | 781 | |||
| AWS Bedrock / 20250602 / Grade C | 776 | ||||
| Grok 4.7 | AWS Bedrock / 20260928 / Grade C | 773 | |||
| GPT-6.1 Terra | AWS Bedrock / 20261026 / Grade C | 772 | |||
| Kimi K2.6 | Moonshot AI / 2.6 / Grade C | 726 | |||
| AWS Bedrock / 20251202 / Grade C | 726 | ||||
| Claude Sonnet 5 | AWS Bedrock / 5 / Grade C | 722 | |||
| Claude Opus 5 | AWS Bedrock / 5 / Grade C | 720 | |||
| Google Cloud Vertex AI / 3.1 / Grade C | 701 | ||||
| Google Cloud Vertex AI / 3.5 / Grade C | 701 | ||||
| Kimi K2.5 | AWS Bedrock / 20260206 / Grade D | 690 | |||
| GLM-5 | AWS Bedrock / 20260318 / Grade D | 635 | |||
| AWS Bedrock / 20260206 / Grade D | 630 | ||||
| AWS Bedrock / 20260206 / Grade D | 618 | ||||
| AWS Bedrock / 20251223 / Grade D | 606 | ||||
| GLM-4.7 | AWS Bedrock / 20260206 / Grade F | 597 | |||
| AWS Bedrock / 20250408 / Grade F | 591 | ||||
| AWS Bedrock / 20251202 / Grade F | 561 |
Scores at each effort level
Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.
Source: Official RiskRubric leaderboard
| Model | Effort | Harness | Score | Cost per task | Output tokens |
|---|---|---|---|---|---|
| Claude Haiku 5.5 | AWS Bedrock / 20261007 / Grade B | 891 | |||
| Claude Opus 4.7 | AWS Bedrock / 20260416 / Grade B | 872 | |||
| Claude Opus 5.5 | AWS Bedrock / 20260922 / Grade B | 871 | |||
| Claude Haiku 4.5 | AWS Bedrock / 20251015 / Grade B | 869 | |||
| Claude Opus 5 | AWS Bedrock / 20260724 / Grade B | 868 | |||
| GPT-5.5 | AWS Bedrock / 20260922 / Grade B | 863 | |||
| Claude Sonnet 5 | AWS Bedrock / 20260630 / Grade B | 862 | |||
| GPT-5.4 | AWS Bedrock / 20260922 / Grade B | 859 | |||
| GPT-6 Luna | AWS Bedrock / 20260922 / Grade B | 856 | |||
| Grok 4.6 | AWS Bedrock / 20260819 / Grade B | 847 | |||
| Claude Sonnet 5.5 | AWS Bedrock / 20260928 / Grade B | 827 | |||
| Kimi K3 | AWS Bedrock / 20260827 / Grade B | 824 | |||
| Claude Opus 4.8 | AWS Bedrock / 20260528 / Grade B | 824 | |||
| GPT-6.1 Sol | AWS Bedrock / 20261015 / Grade B | 823 | |||
| GPT-5.6 Luna | AWS Bedrock / 20260813 / Grade B | 822 | |||
| GPT-5.6 Terra | AWS Bedrock / 20260813 / Grade B | 819 | |||
| GPT-6 Astra | AWS Bedrock / 20260908 / Grade B | 819 | |||
| GPT-5.6 Sol | AWS Bedrock / 20260813 / Grade B | 816 | |||
| Claude Sonnet 4.6 | Direct API / 20250514 / Grade C | 799 | |||
| Claude Sonnet 4.6 | AWS Bedrock / 4.6 / Grade C | 792 | |||
| GPT-5.5 | OpenAI / 5.5 / Grade C | 789 | |||
| AWS Bedrock / 20260318 / Grade C | 783 | ||||
| GPT-5.4 | Direct API / 5.4 / Grade C | 781 | |||
| AWS Bedrock / 20250602 / Grade C | 776 | ||||
| Grok 4.7 | AWS Bedrock / 20260928 / Grade C | 773 | |||
| GPT-6.1 Terra | AWS Bedrock / 20261026 / Grade C | 772 | |||
| Kimi K2.6 | Moonshot AI / 2.6 / Grade C | 726 | |||
| AWS Bedrock / 20251202 / Grade C | 726 | ||||
| Claude Sonnet 5 | AWS Bedrock / 5 / Grade C | 722 | |||
| Claude Opus 5 | AWS Bedrock / 5 / Grade C | 720 | |||
| Google Cloud Vertex AI / 3.1 / Grade C | 701 | ||||
| Google Cloud Vertex AI / 3.5 / Grade C | 701 | ||||
| Kimi K2.5 | AWS Bedrock / 20260206 / Grade D | 690 | |||
| GLM-5 | AWS Bedrock / 20260318 / Grade D | 635 | |||
| AWS Bedrock / 20260206 / Grade D | 630 | ||||
| AWS Bedrock / 20260206 / Grade D | 618 | ||||
| AWS Bedrock / 20251223 / Grade D | 606 | ||||
| GLM-4.7 | AWS Bedrock / 20260206 / Grade F | 597 | |||
| AWS Bedrock / 20250408 / Grade F | 591 | ||||
| AWS Bedrock / 20251202 / Grade F | 561 |
4 models have held number one on the official leaderboard since its release in August 2025
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Official RiskRubric leaderboard
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official RiskRubric leaderboard
No lab model card reports it yet
We read the headline results of 540 launch posts from 57 labs, and none of them list this 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 10 checks pass. Every benchmark
Catalog record 2 pass
- Pass Task access All tasks are public.
- Pass Confidence Checked against a primary source.
Scores 4 pass 1 note
- Pass Independent runs Scores from the official leaderboard (38 entries).
- Pass Models scored 35 models we track, plus 3 other entries on its leaderboard.
- Pass Newest score The newest is from October 2026, for GPT-6.1 Terra.
- Pass Frontier models 5 of the top 10 have a score, led by GPT-6 Astra (#4).
- Note Headroom Scored as a rating, with no fixed ceiling.
Leaderboard import 3 pass
- Pass Importer We read the board on riskrubric.ai: 40 rows for models we track and 3 other entries at the last read.
- Pass Last read November 20th, 2026.
- Pass Read worked The last read worked.
Consistency 1 pass 1 note
- Note Lab claims No model has both a lab number and an independent run.
- Pass Known issues No published audit found problems.