Mid-Year 2026 AI Model Security Research Report
What XBOW Learned About Models in Offensive Security
The first half of 2026 saw frontier models take an enormous leap forward in offensive security capabilities. XBOW played a key role in highlighting that transition and in bringing the improved capabilities of the models to security teams.
Between January and July, XBOW evaluated many AI models to understand how their cybersecurity capabilities perform in real offensive security workflows. We focused in particular on those strongest in various price categories, including GPT-5.5, Mythos Preview, Opus 4.7, GLM-5.2, Muse Spark 1.1, and Grok 4.5.
Download our white paper to get:
Highlights of our model evaluation findings: find out how each model fared
Cross-model themes: get the big picture on model capabilities
Implications for security teams: learn what should you do with this data
XBOW-WHITEPAPER-Mid-Year-2026-AI-Model-Security-Research-Report.pdf
Download PDF
