OpenAI GPT-6 Cyber System Card - SEC-Bench Pro (Highest Tested Budget): leaderboard
Metric: Pass@1 (%). Source: deploymentsafety.openai.com.
Interactive version: theaggregate.ai/benchmark?slug=openai-gpt-6-cyber-system-card-sec-bench-pro-highest-tested-budget · How It Works · Data refreshed daily, snapshot 2026-11-05.