Ant Group’s Robbyant Unveils LingBot-VA 2.0: A Causal Video-Action Model Built Natively for Physical AI
Robbyant, the embodied AI unit inside Ant Group, has released the LingBot-VA 2.0.The first embodied-native foundation model. It describes a video-action foundation model for generalist robot manipulation. The research team pretrains the whole stack for embodiment instead of fine-tuning a video generator. What is LingBot-VA 2.0? Most video-action models reuse two components built for digital content creation. One is a reconstruction-oriented VAE. The other is a bidirectional video-diffusio
Get smart on it
A video-action foundation model called LingBot-VA 2.0 has been released for robot manipulation tasks. Unlike earlier models that adapted components built for digital video creation, this system was designed from the ground up specifically for physical AI, using a causal architecture that unfolds strictly forward in time rather than bidirectionally. The model addresses key limitations of previous approaches by using a semantic visual-action tokenizer that preserves action-relevant information, a sparse mixture-of-experts backbone for efficient processing, and asynchronous execution that allows robots to act while the system plans ahead. In testing on 50 robotic tasks, the system achieved higher performance than comparable models and reached control speeds of 225 Hz through various acceleration techniques.
Related news
Kyutai Releases MuScriptor: An Open-Weight Decoder-Only Transformer for Multi-Instrument Music Transcription to MIDI
Automatic Music Transcription (AMT) converts an audio recording into symbolic notes, usually MIDI. Single-instrument transcription already works reasonably well. However, transcribing a full multi-instrument mix stays difficult. Kyutai and Mirelo team now release MuScriptor to close that gap. It is an open-weight model trained on real, multi-instrument recordings across many genres. This article explains how MuScriptor works, what the benchmarks show, and how to run it. What is MuScriptor
Google Research Introduces SensorFM: A Wearable Health Foundation Model Pretrained on One Trillion Minutes of Sensor Data
Most wearable health models are built one outcome at a time. That approach breaks down at thirty-five endpoints. Labels are expensive and retrospective annotation is infeasible. Google Research introduced SensorFM, a foundation model for wearable health pre-trained on more than 1 trillion minutes of sensor data from 5 million people. https://arxiv.org/pdf/2605.22759 What is SensorFM? SensorFM is a Large Sensor foundation Model for wearable time-series representation learning. It ing
