Embodied AI Model Shows Emergent Generalization; Minimax Video Tested
mishig25 · x · 2026-07-30
- Embodied AI Progress: Physical Intelligence noted that its π0.7 model exhibits emergent compositional generalization, recombining learned skills based on prompts, such as figuring out how to use an air fryer to cook a sweet potato.
- Video Generation Test: A developer used the first frame of this demo as a seed image to test subsequent video generation using Minimax's video model.
More from Embodied
- Neuracore: Fixing the Data Plumbing Bottleneck in Robot Learning — stepjamUK · 2026-07-30
- Gemini Robotics ER 2 in Action: Sub-second Bidirectional Streaming & Multi-Robot Orchestration — _philschmid · 2026-07-30
- Google Unveils Gemini Robotics 2 and ER 2 for Advanced Embodied Reasoning — GoogleAI · 2026-07-30
- DeepMind Details Gemini Robotics 2: Whole-Body Intelligence and Multi-Robot Collaboration — GoogleDeepMind · 2026-07-30
- DeepMind Unveils Three New Models Powering Full-Body Robot Intelligence — GoogleDeepMind · 2026-07-30
- DeepMind Launches Gemini Robotics 2 for Humanoid Whole-Body Control — GoogleDeepMind · 2026-07-30