Benchmarks / Cyber
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.
AA Cyber Index
Higher is better · best effort per model
full table below
| # | Model | Effort | Score |
|---|---|---|---|
| 1 | GPT-6 Sol | max | 69 |
| 2 | Grok 4.7 | xhigh | 56 |
| 3 | MiMo V2.6 Pro | – | 56 |
| 4 | GPT-6 Luna | max | 53 |
| 5 | GLM 5.3 Flash | max | 50 |
| 6 | Mistral Large 4 | – | 50 |
| 7 | Step 5 Preview | – | 48 |
| 8 | Muse Spark 1.3 | xhigh | 44 |
| 9 | Kimi K3 | max | 41 |
| 10 | DeepSeek V4.1 Flash | max | 41 |
| 11 | GLM 5.3 | max | 36 |
| 12 | Beam 1.0 | – | 34 |
| 13 | GPT-6 Astra | max | 33 |
| 14 | Opus 5.5 | max | 29 |
| 15 | Gemini 3.8 Flash | high | 25 |
| 16 | Fable 5.1 | max | 25 |
| 17 | MiniMax M3 | – | 22 |
| 18 | Nemotron 3 Ultra | max | 13 |
| 19 | Qwen 3.8 27B | xhigh | 13 |
| 20 | Qwen 3.8 2.4T-A95B | – | 13 |