Gasgoo Munich- Robbyant, the embodied intelligence arm of Ant Group, has officially launched its first independent funding round. The target is 1.5 billion yuan, according to Gasgoo. The company also plans to pursue a second round by the end of 2026.If this timeline holds, it would set a new fundraising speed record for domestic startups in the embodied intelligence sector.Established in December 2024 and wholly owned by Ant Group, Robbyant positions itself as a foundational model company "building brains for robots." Adhering to an "embodied native" technical approach, it has constructed a comprehensive model ecosystem spanning visual perception, spatial understanding, task decision-making, and motion generation.In Robbyant's view, a sophisticated brain must bridge three physical-world gaps: perception challenges stemming from sensors, high-dynamic interaction, and generalization issues.To address this, Robbyant released the LingBot 1.0 series in late January, covering full-stack technologies from spatial intelligence and dexterous manipulation to environmental interaction. Building on that, it unveiled the next-generation spatial perception model, LingBot-Depth 2.0, in early July, completing a closed-loop capability spanning spatial intelligence, environmental feedback, and dexterous operation.Image Source: RobbyantOn the spatial perception front, the LingBot-Vision foundation model uses geometric structures for visual pre-training, establishing a spatial-native visual base for robots. Building on this, LingBot-Depth 2.0 compensates for sensor errors, specifically tackling real-world challenges like transparent, reflective, and dense objects.Robbyant has partnered with Orbbec to integrate LingBot-Depth 2.0 into its camera products, with these enhanced cameras now available globally.For dynamic interaction, the LingBot-Video model employs a Mixture of Experts (MoE) architecture, delivering systematic improvements in inference efficiency, physical plausibility, action understanding, and task completion. Meanwhile, LingBot-World 2.0 extends continuous generation from minutes to hours and supports 720p/60fps real-time interaction, allowing the environment to evolve continuously based on actions and events.On the action execution front, LingBot-VA 2.0 — the industry's first embodied-native world action model — is trained from scratch. Innovations like a novel semantic vision-action tokenizer, rigorous causal pre-training, MoE architecture, and asynchronous inference enable robots to "deduce, act, and adjust simultaneously."Designed for "one-brain, many-machine" applications, LingBot-VLA 2.0 has been adapted for 17 brands and 20 robot configurations, including Unitree, Agibot, Galbot, and Leju. It extends control across the entire body, covering the head, waist, and chassis.On the commercialization front, Robbyant has deployed solutions in retail and logistics. Notably, its smart pharmacy initiative developed with Sinopharm was named one of the "Top 10 Highlights" at the 2026 World Artificial Intelligence Conference.Ant Group's cumulative investment in Robbyant has already surpassed the total funding raised by top-tier startups in the sector. Following this round, Ant Group will continue to support the unit with AI computing infrastructure and application scenario resources.Building on this foundation, the move to seek independent funding serves two purposes: securing external resources for future R&D, and marking Robbyant's formal transition from internal incubation to independent market-driven growth.