Vals AI Cybersecurity Benchmark: Gap Between Open-Source and Closed-Source Models Narrows
Understand the story
Vals AI has launched the CyberOffense benchmark, which simulates real enterprise network environments, allowing AI agents to independently launch cyberattacks from isolated Kali Linux machines without human assistance, in order to evaluate each model's real offensive and defensive capabilities. On November 7, 2026, The Decoder reported that the latest results from the test show the gap in cyberattack capabilities between open-source and closed-source models continuing to narrow. Public reporting currently focuses mainly on the benchmark's testing methodology and latest results, and the story is still being tracked.
AI overview of reporting · Updated 2 h前
Reporting timeline
Explore different aspects of the story through its reporting.
- The DecoderThe gap between open and closed AI models in cyber capabilities keeps shrinking
Vals AI's CyberOffense benchmark simulates real enterprise networks, where AI agents independently launch attacks from an isolated Kali Linux machine without human assistance.
Story heat trend
Not enough continuous observations to show a trend yet.