TARS Wins SAIL Award at WAIC 2026 with Launch of AWE 3.5 Model

TARS showcased its latest embodied foundation model, AWE 3.5, at the World Artificial Intelligence Conference (WAIC 2026) in Shanghai, where it received the SAIL Award for technical innovation according to a press release. The event took place from July 17 to 20 at the Shanghai World Expo Exhibition and Convention Center.

AWE 3.5 integrates action, perception, geometry, and tactile sensing into a unified system. Built on more than one million hours of human centric data validated in industrial settings, it doubles task execution efficiency compared to its predecessor Pi 0.5. The model supports continuous interaction and introduces an embodied pre training and post training paradigm for scalable iterations.

TARS demonstrated AWE powered robots performing tasks such as phone packing, backpack organization, and precision screw sorting, highlighting the model's ability to generalize across unfamiliar scenarios. The company also recreated an automotive wiring harness production line using multiple A1 robots, showing the technology’s industrial reliability in a smart manufacturing setup.

Another highlight was DexHand’s live performance with magician Deng Nanzi, where the robotic hand mounted on an A1 robot completed actions such as card spreading, handwriting, and solving a Rubik’s Cube. TARS plans to increase its training dataset to 10 million hours by the end of 2026 to further enhance performance and support large scale deployment.