Alibaba's RynnBrain 1.1: Scaling Embodied Foundation Models
Key Questions
What is Alibaba's RynnBrain 1.1?
RynnBrain 1.1 is a family of embodied foundation models from Alibaba scaling from 2B to 122B parameters. It focuses on addressing generalization and data bottleneck challenges in embodied AI.
What new capabilities does RynnBrain 1.1 introduce?
It introduces a unified cross-embodiment action space, contact point prediction, and 3D grounding. These features enhance performance across multiple benchmarks.
How does RynnBrain 1.1 compare to other models?
RynnBrain 1.1 outperforms all proprietary models on multiple benchmarks. The models scale up to 122B parameters while improving cross-embodiment capabilities.
Alibaba released RynnBrain 1.1, a family of embodied foundation models scaling from 2B to 122B parameters. It introduces unified cross-embodiment action space, contact point prediction, and 3D grounding. Outperforms all proprietary models on multiple benchmarks, directly addressing generalization and data bottleneck challenges.