
FRONTIER LABS
Xenon Releases Hunmin VLM 397B for Autonomous Computer Operation
The new 397-billion parameter model utilizes FP8 quantization and specialized reinforcement learning to achieve high-accuracy screen recognition and system control.

The new 397-billion parameter model utilizes FP8 quantization and specialized reinforcement learning to achieve high-accuracy screen recognition and system control.

The new open-source model enhances AV decision-making by integrating high-level reasoning with 360-degree sensor fusion.

This powerful new reasoning model, developed after a major internal overhaul, marks a proprietary shift aimed at deep integration across the company’s product ecosystem.
Applied machine learning, filed daily.
Model releases, silicon, clinical deployment, and the policy shaping them. No digest padding.