# Artificial Analysis Cyber Index (AA Cyber Index)

> Artificial Analysis' composite index of cyber-defense capability: finding vulnerabilities in real codebases, reproducing and validating them, and patching them without breaking existing functionality. It averages CWE-Bench-AA, DeepsecBench-AA and CyberGym-E2E-AA.

Canonical page: https://benchlm.ai/benchmarks/aacyberindex

- Category: [Agentic](/agentic)
- Last updated: October 29, 2026

## About AA Cyber Index

- Year: 2026
- Tasks: Three defensive cybersecurity evaluations
- Format: Index score (0-100)
- Difficulty: Real-world vulnerability discovery and patching
- Paper: [Artificial Analysis Cyber Index](https://artificialanalysis.ai/evaluations/artificial-analysis-cyber-index)

Cyber Index v1, launched October 2026. Independently run by Artificial Analysis under its published harness. Display-only; not yet admitted as independent-run evidence.

AA Cyber Index is currently displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.

## Leaderboard (19 models)

| Rank | Model | Creator | Score |
|------|-------|---------|-------|
| 1 | [Kimi K3.1 Pro](/models/kimi-3-1-pro) | Moonshot AI | 64.1% |
| 2 | [Grok 4.7](/models/grok-4-7) | xAI | 56.4% |
| 3 | [MiMo-V2.6-Pro](/models/mimo-v2-6-pro) | Xiaomi | 56.1% |
| 4 | [GPT-6 Luna](/models/gpt-6-luna) | OpenAI | 52.7% |
| 5 | [GLM-5.3-Flash](/models/glm-5-3-flash) | Z.AI | 49.9% |
| 6 | [Mistral Large 4](/models/mistral-large-4) | Mistral | 49.5% |
| 7 | [Step 5 Preview](/models/step-5-preview) | StepFun | 47.7% |
| 8 | [Kimi K3](/models/kimi-k3) | Moonshot AI | 41.3% |
| 9 | [DeepSeek V4.1 Flash 1017](/models/deepseek-v4-1-flash) | DeepSeek | 40.8% |
| 10 | [GPT-6 Sol](/models/gpt-6-sol) | OpenAI | 36.8% |
| 11 | [GLM-5.3](/models/glm-5-3) | Z.AI | 36.4% |
| 12 | [Beam 1.0 501B-A23B](/models/reflection-beam) | Reflection AI | 33.7% |
| 13 | [GPT-6 Astra](/models/gpt-6-astra) | OpenAI | 33.4% |
| 14 | [Claude Opus 5.5](/models/claude-opus-5-5) | Anthropic | 28.7% |
| 15 | [Gemini 3.8 Flash](/models/gemini-3-8-flash) | Google | 25.1% |
| 16 | [Claude Fable 5.1](/models/claude-fable-5-1) | Anthropic | 25.0% |
| 17 | [MiniMax M3](/models/minimax-m3) | MiniMax | 22.4% |
| 18 | [Nemotron 3 Ultra](/models/nemotron-3-ultra) | NVIDIA | 13.1% |
| 19 | [Qwen3.8-27B](/models/qwen3-8-27b) | Alibaba | 12.7% |

## FAQ

### What does AA Cyber Index measure?

Artificial Analysis' composite index of cyber-defense capability: finding vulnerabilities in real codebases, reproducing and validating them, and patching them without breaking existing functionality. It averages CWE-Bench-AA, DeepsecBench-AA and CyberGym-E2E-AA.

### Which model scores highest on AA Cyber Index?

Kimi K3.1 Pro by Moonshot AI currently leads with a score of 64.1% on AA Cyber Index.

### How many models are evaluated on AA Cyber Index?

19 AI models have published results on AA Cyber Index in the BenchLM catalog.

### Does AA Cyber Index affect BenchLM's overall score?

Not directly. AA Cyber Index is still displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.

## Compare Top Models on AA Cyber Index

- [Kimi K3.1 Pro vs Grok 4.7](/compare/grok-4-7-vs-kimi-3-1-pro)
- [Grok 4.7 vs MiMo-V2.6-Pro](/compare/grok-4-7-vs-mimo-v2-6-pro)
- [MiMo-V2.6-Pro vs GPT-6 Luna](/compare/gpt-6-luna-vs-mimo-v2-6-pro)
- [GPT-6 Luna vs GLM-5.3-Flash](/compare/glm-5-3-flash-vs-gpt-6-luna)