- Li Auto introduced the Mach M100 AI chip with 1,280 TOPS computing power and over 82% real-world efficiency.
- Halo OS and upcoming OTA releases aim to enhance assisted driving, automation, and human-like vehicle intelligence.
Li Auto hosted its Livis Day software and embodied intelligence launch event on June 15, presenting a new vision for intelligent mobility. The company described an embodied intelligent vehicle as a fusion of an electric vehicle, a professional driver, an AI computer, and a life assistant. As part of this strategy, the automaker introduced its first dynamic dataflow AI chip, the Mach M100, alongside a new software foundation designed to support advanced vehicle intelligence and automation capabilities.
Mach M100 AI Chip Debuts with High Computing Performance
The newly introduced Mach M100 is built using a 5 nm automotive-grade manufacturing process and delivers 1,280 TOPS of computing power from a single chip. The processor adopts a dataflow architecture that enables more than 82% real-world computational efficiency. The architecture is designed to support demanding artificial intelligence workloads required for next-generation intelligent driving, perception systems, decision-making functions, and vehicle control applications.
Halo OS Enables End-to-End Embodied Intelligence
Built around the Mach M100 platform, Li Auto has also developed Halo OS, a dedicated embodied intelligence operating system. The platform integrates the complete perception, decision, and execution chain into a unified architecture. Combined with a fully drive-by-wire chassis capable of ultra-fast response, the system reduces end-to-end latency to 0.28 seconds. According to the company, this response time is approximately 40% faster than typical human reaction speed, supporting more responsive and intelligent vehicle behavior.
Key Specifications of the Mach M100 Platform
| Feature | Specification |
|---|---|
| Manufacturing Process | 5 nm Automotive-Grade |
| Computing Power | 1,280 TOPS |
| Architecture | Dynamic Dataflow |
| Computational Efficiency | Over 82% |
| End-to-End Latency | 0.28 Seconds |
OTA Roadmap Targets Major Capability Expansion in 2026
Li Auto outlined a phased software roadmap for 2026 through a series of over-the-air updates. The July release is expected to improve assisted-driving efficiency by approximately 30%. Additional functions will include a convenient tire-change capability supported by active suspension technology and a map-based travel Agent designed to improve navigation and trip planning experiences.
The September update will introduce more human-like system capabilities, enabling vehicles to independently manage complex situations. Planned features include full-scenario automated reversing and road-surface learning functionality. During the same phase, the Agent will gain connectivity with users’ personal computers and smartphones. Support for Apple CarPlay and additional smartphone-to-vehicle integration functions is also scheduled to become available.
By December, Li Auto intends to further improve the safety and efficiency of Livis technology to exceed human-level performance benchmarks. The company stated that reaction speed is expected to become 56% faster than human response capability. Additional features will include exterior user recognition for automatic profile switching and the ability to interpret and respond to traffic police hand signals, expanding real-world situational awareness.
Embodied Intelligence Becomes a Core Development Direction
The combination of the Mach M100 chip, Halo OS, drive-by-wire technology, and a structured OTA deployment strategy reflects Li Auto’s focus on embodied intelligence as a central development path. By integrating computing, perception, decision-making, and execution into a unified platform, the company aims to enhance automation, improve safety performance, and deliver more natural interactions between vehicles and users through future software-defined mobility systems.
Frequently Asked Questions
What is the Mach M100 chip introduced by Li Auto?
The Mach M100 is Li Auto’s first dynamic dataflow AI chip developed for embodied intelligent vehicles. It is manufactured using a 5 nm automotive-grade process and delivers 1,280 TOPS of computing power. The chip uses a dataflow architecture designed to achieve more than 82% real-world computational efficiency. It serves as the hardware foundation for advanced perception, decision-making, intelligent driving, and vehicle control functions within the company’s future software-defined vehicle ecosystem.
What are the main capabilities planned in Li Auto’s 2026 OTA updates?
Li Auto’s 2026 OTA roadmap includes multiple stages of capability expansion focused on intelligent driving and user experience. The July update aims to improve assisted-driving efficiency and introduce new convenience features. September will add human-like capabilities such as automated reversing and road-surface learning, along with expanded device connectivity. By December, the company plans to enhance safety and efficiency further, enable automatic profile switching through exterior recognition, and add traffic police hand-signal recognition functionality.
Top of Form
Bottom of Form
Click above to visit the official source.