NIO launches NOMI GPT large model

Shanghai (Gasgoo)- On April 12, Chinese electric vehicle maker NIO officially rolled out its NOMI GPT large model, which has been deployed onto the vehicles equipped with the NIO Banyan intelligent system, according to a post on NIO's WeChat account. 

The NOMI GPT device-to-cloud multimodal large model newly launched today is tailored to power NIO's NOMI digital in-car system. Encompassing NIO’s in-house developed multimodal perception, cognitive center, emotional engine, and multiple AI-enabled agents, it seamlessly integrates NIO's products, services and community to provide more efficient and enjoyable AI services.

According to NIO, this latest upgrade to NOMI involves numerous new features based on large model capabilities. Currently, NOMI can answer encyclopedic questions  across various domains such as science, history, and entertainment, and is able to conduct continuous lively dialougues. Additionally, NOMI can generate AI driving scenarios, ambient light combinations, and unique expressions based on simple commands.

The upgraded NOMI now possesses capabilities such as "seeing and recognizing" "hearing and understanding" and "sensing temperature and touch" while also ensuring account isolation to prevent sensitive information within the car. End-to-end encryption of in-car communication guarantees users’ privacy and security. 

According to the automaker, NIO's self-developed device-side multimodal perception boasts billions of parameters, integrating various modalities including image (visual), sound (audio), and vehicle sensors (tactile). It can describe multidimensional perception in natural language just like human, thus perceiving both the interior and exterior of the vehicle, and its surroundings in a more comprehensive way. 

It is noteworthy that the internal testing of the device-side multimodal perception is currently underway, and the feature is set for application in cars later this year.

Thanks to the upgrade, NOMI will possess enhanced cognitive abilities, enabling precise determination of user intentions and needs from vast information sources. NIO's self-developed cognitive center can leverage multidimensional perceptual information to form complex task processing capabilities through inferential analysis. This cognitive center integrates the most comprehensive industry knowledge base, adeptly managing various tasks and answering diverse questions, including distribution, multimodal rejection, inheritance and jump, and multi-agent arbitration.

Furthermore, NOMI upgraded will feature an exclusive emotion engine, which serves as the key to achieving humanoid interaction. It will respond to users’ requests with millisecond-level speed in aspects including perception, cognition, decision-making, and so on.

The upgraded NOMI will also come with an interactive window. NIO has developed a brand-new interactive interface for NOMI GPT, facilitating more dynamic presentation of various types of generated content. Additionally, NOMI's improved knowledge base and reasoning capabilities enables it to answer users’questions related to encyclopedic knowledge, travel scenario, and personal circumstances.

Moreover, NOMI also supports users in searching vehicle information, vehicle knowledge, indicators, etc., assisting both new and existing users in better understanding and use their vehicles.