340 billion parameter model, AutoBrains and NVIDIA expand collaboration

Autobrains said it will use the inference capabilities of NVIDIA's 340-billion-parameter Alpamayo 2 Super autonomous-driving model to build agents tailored to specific driving situations, expanding a partnership that previously focused on automotive platforms. The company will transfer the foundation model's knowledge into its own agentic AI architecture to create agents that respond to defined driving scenarios. Alpamayo 2 Super is a reasoning-capable vision-language-action model that generates driving trajectories and causal reasoning records for complex situations. The move matters because it pairs NVIDIA's large-scale driving foundation model with Autobrains' agent-based stack, signaling a shift toward modular, situation-specific autonomy. What to watch next is how quickly Autobrains converts the model's knowledge into production-ready agents and whether the expanded collaboration extends beyond development into deployed vehicle programs.