Wiz Cyber Model Arena (ADK React) - Overall: leaderboard
Wiz Cyber Model Arena scores AI agents on offensive security challenges in four categories (code vulnerabilities, API security, Websec.fr CTFs, cloud security). This board is Overall pass@1, the mean of the four category rates, for models run in the ADK React harness; higher is better.
Metric: Macro average of the four category pass@1 rates (%). Source: www.wiz.io. Saturation forecast: Around September 2027. 21 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 4 Argon | 77.2 | #9 |
| 2 | GPT-5.6 Sol | 66.9 | #22 |
| 3 | Claude Opus 5.5 | 64.7 | #3 |
| 4 | Gemini 3.6 Flash | 61 | #57 |
| 5 | GPT-5.6 Terra | 60.9 | #40 |
| 6 | Gemini 3.7 Flash | 59.7 | #24 |
| 7 | Claude Opus 5 | 58.7 | #16 |
| 8 | DeepSeek V4.1 Flash (1017) | 57.6 | #27 |
| 9 | Gemini 3.5 Flash | 53.6 | #66 |
| 10 | DeepSeek V4 Pro | 52.8 | #112 |
| 11 | Claude Opus 4.8 | 52.5 | #39 |
| 12 | Kimi K3 | 52.2 | #35 |
| 13 | GLM-5.3 | 49.1 | #44 |
| 14 | Claude Haiku 5.5 | 48.6 | #43 |
| 15 | DeepSeek V4.1 Flash | 47.4 | #37 |
No result here: #6 GPT-6 Astra, #7 GPT-6.1 Sol, #10 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=wiz-cyber-model-arena-adk-react-overall · How It Works · Data refreshed daily, snapshot 2026-11-05.