Vals Finds Open AI Models Closing the Hacking Gap with Gated Frontier Systems

Free Chinese AI Models Are Getting Scary Good at Hacking

An experiment in which AI models were asked to break into simulated networks showed the hacking skills of DeepSeek and GLM to be not far behind OpenAI’s most advanced cyber model, which isn’t public.

Vals finds open AI models closing the hacking gap with gated frontier systems

Vals AI's CyberOffense benchmark shows open models like DeepSeek V4.1 nearing gated GPT-6 Sol on hacking tasks at a fraction of the cost.