Xiaomi Open Sources Full Post-Training Process for VLA Large Model, Enabling Robotic Sub-Millimeter-Level Operations

Xiaomi has once again dropped a "heavy bomb" in the field of robot technology. Following the release and open-sourcing of its VLA large model Xiaomi-Robotics-0 in February this year, Xiaomi officially announced the full post-training process of the model today. This move aims to solve the "last mile" problem of robots transitioning from laboratories to practical production, making AI robots truly productive tools that are ready to use out of the box.

In the latest capability demonstration, the robot equipped with this model showcased astonishing fine operation capabilities. Through only 20 hours of task data reinforcement training, Xiaomi-Robotics-0 successfully mastered the highly challenging action of "continuously organizing earphones." In the video, the robot's movements are smooth, accurately placing small earphones one by one into compact storage slots.

This seemingly simple action actually hides technical intricacies. Official technical information indicates that this task faces two core technical barriers: first, the precision challenge. The tolerance between the earphone and the charging case slot is extremely small, requiring the model to have sub-millimeter-level spatial perception capabilities to achieve accurate alignment; second, the stability challenge. Due to the extremely smooth surface of the earphone and the case (with a roughness as low as Ra 0.03μm), it is prone to slipping during contact. The model must have high real-time feedback and action correction capabilities to prevent assembly failure.

Notably, Xiaomi-Robotics-0 ranked in the top six of the global VLA model download list on HuggingFace within its first month of release, showing high industry attention. To further promote the development of the developer ecosystem, Xiaomi has now fully opened up the project's technical report, model weights, and source code.

Currently, developers can access relevant resources through the Xiaomi Robot Technology website and open-source platforms such as GitHub. With the open-sourcing of the full post-training process, the industry generally expects this will significantly lower the development threshold for high-performance robot tasks and accelerate the application of intelligent robots in complex and precise scenarios.

Technical website: https://robotics.xiaomi.com/xiaomi-robotics-0.html#pack-earbuds

Sub-millimeter Precision Alignment: Xiaomi Open Sources the Full Post-Training Process of VLA Large Model

Xiaomi has recently open-sourced the real-world post-training process of its vision-language-action large model, Xiaomi-Robotics-0, promoting the development of embodied intelligence. The team enabled the robot to master precise earphone storage and other complex tasks using only about 20 hours of task data, demonstrating the ability to quickly learn complex skills.

Xiaomi Launches the Most Powerful Model Series MiMo-V2.5, Official Public Testing Begins

Xiaomi released the MiMo-V2.5 series of large models on April 23 and initiated public testing. The series includes four models, with the core models MiMo-V2.5-Pro and MiMo-V2.5 being open-sourced globally, demonstrating its commitment to promoting an open AI ecosystem. This update is not only a product iteration but also a comprehensive upgrade of the technology foundation, featuring flagship performance that supports a context length of up to one million and complex task processing.

Xiaomi MiMo-V2.5 Shocking Beta Test: 4.3 Hours of Manual Compiler Development, Long-Range Intelligent Agent Achieves a Full Leap

Xiaomi released the MiMo-V2.5 series of large models, including MiMo-V2.5, V2.5-Pro, and accompanying TTS and ASR models, marking an upgrade from "usable" to "user-friendly." The flagship model MiMo-V2.5-Pro has reached competitive levels with top models such as Claude Opus4.6 and GPT-5.4 in terms of general intelligent agent capabilities and software engineering. Its core advantage lies in high instruction adherence and self-correction capabilities.

The Explosion of the Domestic Agent Ecosystem! Xiaomi MiMo-V2 Joins the Top Framework Hermes and Opens a 14-Day Free Trial

Xiaomi's self-developed large model MiMo-V2 series has officially joined the globally top-tier open-source Agent framework Hermes Agent, achieving a strong combination. Developers can directly call Xiaomi's flagship model through Nous Portal after updating the framework. At the same time, Xiaomi has launched a two-week free trial "family pack" activity to reward developers.

Xiaomi Open Sources Major Project! OmniVoice Covers 600+ Languages for Zero-Shot Speech Cloning TTS: WER Only 0.84%, 40 Times Faster, Small Languages Can Also Be Resurrected Easily

Xiaomi Kaldi team open-sources the OmniVoice model, supporting over 600 languages. It achieves SOTA performance in multiple metrics on Chinese and multilingual TTS benchmark tests. The Chinese WER is as low as 0.84%, and the multilingual performance surpasses mainstream commercial models, achieving a new breakthrough in speech synthesis.