We're excited to finally release our next generation model for embodied intelligence. We packed a lot into this release, from new architectures (2-4x speed up) to adding audio as a first class modality to new post-training, we're building the foundations for anyone deploying anything in the physical world.