Benchmarks · safety
AA Cyber Index
Artificial Analysis Cyber Index: cyber defense in real repositories (find, reproduce and patch vulnerabilities), the equal-weight mean of CWE-Bench-AA, DeepsecBench-AA and CyberGym-E2E-AA. Safety refusals score zero. Includes trusted-access models that are not publicly available.
17 models · 17 results · data as of 2026-10-29 · artificialanalysis.ai ↗
Leaderboard
DeepSeek V4.1 Flash 1017 max
All results
17 rows| # | Model | Effort | AA Cyber Index |
|---|---|---|---|
| 1 | DeepSeek V4.1 Flash 1017 | max | 56.8 |
| 2 | Grok 4.7 | xhigh | 56.4 |
| 3 | MiMo V2.6 Pro | 56.1 | |
| 4 | GPT-6 Luna | max | 52.7 |
| 5 | GLM 5.3 Flash | max | 49.9 |
| 6 | Muse Spark 1.3 | xhigh | 44.3 |
| 7 | Kimi K3 | max | 41.3 |
| 8 | DeepSeek V4.1 Flash | max | 40.8 |
| 9 | GPT-6 Sol | max | 36.8 |
| 10 | GLM 5.3 | max | 36.4 |
| 11 | GPT-6 Astra | max | 33.4 |
| 12 | Opus 5.5 | max | 28.7 |
| 13 | Gemini 3.8 Flash | high | 25.1 |
| 14 | Fable 5.1 | max | 25.0 |
| 15 | MiniMax M3 | 22.4 | |
| 16 | Qwen 3.8 27B | xhigh | 12.7 |
| 17 | Qwen 3.8 2.4T-A95B | 12.6 |