Xiaomi's AI Ambitions Gain Momentum with Model Adoption and Robotics Wins

Xiaomi Group's sustained investment in artificial intelligence is yielding significant, measurable returns. The company's core base model, Xiaomi MiMo-V2.5, has seen substantial global adoption since its release. Its performance on leading AI infrastructure platforms now highlights its growing influence in the developer community.

Leading Platform Adoption Signals Robust Performance

MiMo-V2.5 recently achieved the top position for weekly API call volume on OpenRouter, a prominent platform for AI model inference. The model processed a total of 10.5 trillion tokens during the measured period. This leading position goes beyond ranking; it serves as a strong market indicator of the model's reliability, efficiency, and utility in real-world applications. The high call volume suggests developers are actively integrating it into diverse products, affirming Xiaomi's competitive footing in foundational AI technology.

Ecosystem Expansion: From Model to Integrated Solutions

Building on the model's success, Xiaomi is deepening its AI integration into practical domains. In June, the company launched Xiaomi Miloco 2.0, an open-source, whole-home AI solution architected around the MiMo model. This framework aims to bridge powerful general AI capabilities with specific smart home devices and interactive scenarios, offering developers a customizable platform for creating context-aware home intelligence. This move represents Xiaomi's push to embed AI directly into the fabric of daily living.

Robotics Breakthroughs Showcase Embodied Intelligence Prowess

While software achievements are critical, Xiaomi's recent victories in international robotics competitions demonstrate its advanced work in embodied AI—where intelligence meets physical action. Also in June, the Xiaomi Robotics team secured top honors at two premier conferences:

  • CVPR 2026 Workshops: The team won first place in the GigaBrain Challenge-RoboChallenge track, which focuses on large-scale perception and decision-making for robots.
  • ICRA 2026: The team clinched another championship in a World Robot Conference (WBC) track, highlighting strengths in physical control and real-world execution.

These dual victories provide concrete evidence of the team's ability to effectively integrate AI models with robotic platforms, advancing systems that can both reason and interact with the physical environment.

Taken together, the widespread adoption of its base model, the launch of an open ecosystem for home AI, and validated breakthroughs in robotics illustrate a multi-faceted AI strategy for Xiaomi. The company appears to be transitioning from technological development to simultaneous execution and ecosystem building.