Alibaba DAMO Academy open-sources embodied brain base model RynnBrain; robots gain spatial-temporal memory for the first time.

```

On February 10, Alibaba DAMO Academy released the embodied AI foundational model RynnBrain, and open-sourced a full series of seven models at once, including a 30B MoE. RynnBrain enables robots to possess spatiotemporal memory and spatial reasoning abilities for the first time, achieving a significant leap in intelligence. It set new records (SOTA) across 16 embodied open-source evaluation rankings, surpassing industry-leading models like Google's Gemini Robotics ER 1.5.

Reportedly, DAMO Academy’s RynnBrain model has creatively introduced spatiotemporal memory and physical world reasoning, which are two fundamental abilities required for robots to interact with their environment. For example, a robot running the RynnBrain model, if interrupted while performing task A and asked to do task B first, can remember the temporal and spatial status of task A and resume working after completing task B.

RynnBrain also has good scalability, enabling rapid training for various embodied models such as navigation, planning, and actions, and is expected to become the foundational model for the embodied industry. For example, embodied planning models require strong predictive and scene analysis abilities, but with RynnBrain, only a few hundred data samples are needed for fine-tuning to surpass Gemini 3 Pro and easily achieve SOTA.

DAMO Academy has now open-sourced the complete series of RynnBrain models, totaling seven, including full-scale foundation models and post-training proprietary models. Among them is the industry’s first 30B embodied MoE architecture model, making robot actions faster and smoother. DAMO Academy also open-sourced the new evaluation benchmark RynnBrain-Bench for assessing spatiotemporal fine-grained embodied tasks, filling a gap in the industry.

Zhao Deli, head of DAMO Academy’s Embodied Intelligence Laboratory, stated: “RynnBrain has achieved, for the first time, the brain’s deep understanding of the physical world and reliable planning, marking a key step towards general embodied intelligence under a hierarchical brain architecture. We expect it to accelerate AI’s deployment from the digital world towards real physical scenarios.”

According to reports, DAMO Academy is actively investing in embodied intelligence and is building deployable, scalable, and evolvable embodied intelligence systems. It has already open-sourced embodied models like WorldVLA, which integrates world models and VLA models, and RynnEC, a world understanding model, as well as the industry’s first robot context protocol, RynnRCP.

Source: Shanghai Securities News

Risk Warning and DisclaimerThe market carries risks, and investments should be made cautiously. This article does not constitute personal investment advice, nor does it take into account the special investment objectives, financial situation, or needs of individual users. Users should consider whether any opinions, viewpoints, or conclusions in this article are suitable for their particular situation. Investing accordingly is at your own risk. ```