At the recent Livis Day Li Auto Software and Embodied Intelligence Conference, Li Auto formally presented its comprehensive definition for an embodied intelligent vehicle. This definition characterizes such a vehicle as an intelligent entity possessing four core capabilities: those of an electric vehicle, a professional driver, an AI computer, and a life assistant. The event systematically showcased a new generation of Li Auto's spatial interaction experience, its self-developed embodied AI model, the world's first self-developed dynamic data flow AI chip—the Mahe M100, along with related products and experiences. The company also announced a full-year OTA upgrade roadmap.

Li Auto's Chairman and CEO, Li Xiang, stated: "What is the relationship between embodied intelligence and car manufacturing? Today's traditional smart cars are not truly intelligent; they are essentially 'function-driven,' serving specific scenarios, and their definition of safety is also 'functional safety.' A genuine embodied intelligent vehicle is one that can protect human safety, complete tasks independently, and be more efficient than a human. This is Li Auto's answer. Over the past decade, we created a mobile home. In the next decade, we will give life to both the car and the home. This is the purpose of our company's existence, and we hope to witness this new era together with everyone!"

The new-generation flagship cockpit from Li Auto features the world's first Qualcomm 8797 Elite flagship chip. It is equipped with the industry's first ultra-wide panoramic screen boasting a 90Hz high refresh rate and 180Hz touch sampling rate. The cockpit also incorporates a 9.3.6 Flawless Starlight Theater audio system with a peak power output of 5440W. Furthermore, the new cockpit supports CarPlay and connectivity with all mainstream flagship smartphones, with CarPlay supporting synchronized HUD display for maps and steering wheel control.

The brain of the embodied intelligence system is formed by the deep integration of linguistic intelligence and machine intelligence. Linguistic intelligence is powered by Li Auto's self-developed Mahe Mind-Pro and Mahe Mind-Edge, responsible for language and logical reasoning, understanding commands, and deducing action plans. Machine intelligence is powered by the self-developed Mahe VLA, responsible for 3D visual perception and physical motion control, perceiving the physical world clearly and executing operations precisely. As of June 14, 2026, the intelligent driving assistance systems across all Li Auto models have cumulatively performed over 17.27 million proactive hazard avoidances, with 55,671 significant avoidances. From 2022 to the present, the daily average for proactive hazard avoidance is 12,000 times. This year, the dual-Mahe intelligent driving assistance system has added recognition of suspended obstacles and increased the nighttime emergency braking limit to 130 km/h. Zhan Kun, Head of Base Models at Li Auto, indicated that the capabilities of Mahe VLA will continue to evolve in the second half of the year, with a new Mahe VLA model being pushed to AD Max in the third quarter and capabilities aligning with Tesla's FSD V14 by the fourth quarter.

Beginning with its formal project initiation in 2022 and after three and a half years of continuous refinement, Li Auto's self-developed Mahe M100 chip achieved mass production and vehicle integration in May 2026, becoming the world's first mass-produced dynamic data flow AI chip. With the mass deployment of the Mahe M100, Li Auto has achieved full-stack self-research encompassing the chip, compiler, operating system, AI algorithms, and domain controllers. Xie Yan, CTO of Li Auto, stated that the Mahe M100 employs a 5-nanometer automotive-grade process, delivers a single-chip computing power of 1280 TOPS, is currently the world's most powerful mass-produced automotive-grade inference chip, and achieves an actual operational efficiency exceeding 82%.

Concluding the event, Li Xiang announced the OTA plan for the Livis model. It is reported that in 2026, the Livis embodied intelligence system will reach key development milestones in July, September, and December. The July theme is "Overall 30% Improvement in Intelligent Driving Efficiency," assisting users through complex scenarios like width-restriction piers and height-restriction bars, officially launching travel guide-related Agent skills, while adding features such as an intercom for car owners, a sentry mode consuming two kilowatt-hours of electricity per day, and convenient tire changing. The September theme is "Reversing Like a Human," achieving full-scenario autonomous reverse driving capabilities including navigating narrow roads and yielding while reversing, active suspension handling complex road surfaces, intelligent control of parking locks and garage doors, Agent connectivity to computers and phones, and the addition of Super CarPlay. The December theme is "Safety and Efficiency Surpassing Humans," adding capabilities such as active protection against unintended steering wheel contact, emergency evasion compensation, obeying traffic police commands, and Face ID recognition for account switching. The end-to-end response speed will reach 0.2 seconds, 56% faster than a human, comprehensively surpassing human capabilities.