ItShiZhiHang has unveiled its new embodied native foundation model, AWE 3.5, at the 2026 WAIC. This model is designed to handle multiple tasks across various industrial scenarios, moving beyond its initial success in wire harness assembly. AWE 3.5 integrates vision, language, and action modalities from the pre-training phase, enabling it to act as both a policy generator and a world model simulator. This approach allows for rapid iteration and reinforcement learning within the model itself, reducing the need for extensive real-world robotic testing. AI
IMPACT This release signifies a step towards more capable and adaptable embodied AI systems, potentially accelerating their integration into complex industrial environments.
RANK_REASON New embodied native foundation model release from a significant AI lab. [lever_c_demoted from frontier_release: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
