Atlas

Benchmarks

← All benchmarks

GDPx: Cybersecurity

Cybersecurity · 2026-11-01

GDPx's Cybersecurity domain: 18 tasks from Arithmetic on finding, exploiting and fixing proprietary zero-days inside live, multi-service enterprise environments. Scores are pass@1 over three attempts per task. A component of the GDPx score, which weights its 11 domains equally.

Top models (higher is better)

ModelScore
Muse Wonder Full35.2
Muse Wonder24.1
Gemini 4 Argon18.5
Gemini 4 Carbon16.7
Grok 4.714.8
GPT-6 Luna11.1
Kimi K311.1
Claude Haiku 5.59.3
Claude Opus 5.59.3
GLM-5.37.4
MiniMax M3.1 Flash7.4
Claude Fable 5.13.7
Claude Sonnet 5.53.7
GPT-6.1 Sol0.0
GPT-6.1 Terra0.0
GPT-6 Astra0.0
Muse Spark Open0.0
Loading Atlas data…