Loading...
Loading...
Xiaomi has made a major open-source push in autonomous driving with OneVL, a unified latent-space model that merges vision-language-action (VLA), world models, and latent reasoning into one architecture. Published with weights, code and an arXiv technical report, OneVL claims state-of-the-art results on several driving benchmarks while improving speed and decision clarity via XLA-style inference. The release arrives as Xiaomi Auto reorganizes leadership—appointing former Tesla Shanghai factory chief Song Gang to run production and smart manufacturing—signaling the company’s move to scale both software and vehicle manufacturing. Parallel gains from rivals like XPeng’s VLA rollout highlight intensifying competition in driving AI and ADAS adoption.
Xiaomi open-sourcing OneVL accelerates access to advanced unified VLA and world-model techniques for autonomous driving, affecting how teams build and test driving AI. Leadership changes to boost production and smart manufacturing signal Xiaomi's intent to scale software into vehicles, affecting integration and deployment timelines.
Dossier last updated: 2026-05-20 01:46:04
Xiaomi announced and open-sourced Xiaomi OneVL, a unified latent-space language-visual-autonomous-driving framework that, Xiaomi says, is the industry first to integrate VLA (vision-language-action), world models, and latent-space reasoning into one architecture. Founder and CEO Lei Jun highlighted that OneVL sets new performance ceilings on multiple inference and planning benchmarks, improving both speed and accuracy versus explicit chain-of-thought and matching “answer-only” latent CoT methods. Xiaomi published the technical report on arXiv, released model weights, and open-sourced training and inference code, inviting developers and researchers to explore applications in autonomous driving and embodied intelligence. The release signals a major open-source push from Xiaomi into driving-scale multimodal models.
Xiaomi released and open-sourced Xiaomi OneVL, a unified driving model that for the first time combines VLA (visual-language-action), world models, and latent-space reasoning into a single framework. Xiaomi says OneVL leverages XLA-style inference to boost speed and accuracy, surpassing explicit chain-of-thought approaches in precision while matching the speed of “answer-only” latent CoT methods. The model reportedly achieves SOTA results on ROADWork, Impromptu and Alpamayo-R1 benchmarks and strong performance on NAVSIM, and offers both textual and visual explanations for decisions. Xiaomi published weights, training and inference code, plus a technical report and project site on GitHub and arXiv, enabling researchers and practitioners to adopt and extend the system.
XPeng said on May 8 that after a month-long rollout of its second-generation VLA (VIA) intelligent driving system, assisted-driving mileage powered by the system has exceeded 50% of total miles. The company reported a 25.87% month-on-month drop in takeovers per 100 km and a 27.84% rise in trips completed entirely with VLA. During the May Day holiday, daily AI-assist usage hit 93.21%, cumulative assisted miles reached 84.46 million km, and the longest single-vehicle assisted run was 5,441 km. The metrics suggest improved system stability and higher real-world adoption for XPeng’s in-house driving stack. This matters for competition in autonomous and driver-assist features among EV makers and ADAS suppliers.
Xiaomi Auto has reassigned Song Gang to oversee production, smart manufacturing and systems operations, reporting directly to CEO and Auto Division president Lei Jun. Song, who joined Xiaomi in April as a vice president and chief strategist for the auto unit, previously spent years at Tesla Shanghai where he served as senior director and vice president of manufacturing and as Shanghai Gigafactory plant director, and led establishment of Tesla’s Shanghai energy storage factory. He also worked at Envision Energy as SVP of integrated supply chain in 2024. The move brings experienced EV manufacturing leadership to Xiaomi as it scales vehicle production and industrializes smart manufacturing.