AgiBot WITA-Omni Full-Modal Model Tops DailyOmni Global Leaderboard: Beating Google Gemini, ByteDance Doubao, and Alibaba Qwen at Embodied Cross-Modal Understanding

AgiBot WITA-Omni scores 85.21 on DailyOmni benchmark, 6 of 8 indicators first place, using Thinker-Talker-Actor architecture that synchronizes speech, action, and expression on a single timeline.

Read the full article at the source →