Nio will launch its Banyan 3.1.0 system in January, for the first time to bring smart driving capabilities that incorporate end-to-end AI models, according to local media.

Nio Inc (NYSE: NIO) plans to launch its Banyan 3.1.0 system in January 2025, for the first time to bring smart driving features that incorporate end-to-end artificial intelligence (AI) large model architecture, according to a report in local media outlet AutoPix today.

The Nio brand currently delivers eight models in China -- ES8, ES7, ES6, EC7, EC6, ET7, ET5, ET5 Touring -- and they're all based on the NT 2.0 technology platform, with the vehicle system being Banyan.

The brand's earlier models were built on the NT 1.0 technology platform, with the vehicle system being Aspen. The ET9, unveiled last December, is based on the next-generation technology platform, NT 3.0, and deliveries are expected to begin in the first quarter of 2025.

At the end of August, Nio launched the Banyan 3.0.0 system, focusing on upgrading the smart cockpit, smart chassis and incorporating AI technologies, which is the latest version of the current Banyan system.

Nio's development process for the end-to-end large model hasn't been a smooth one, and its local counterparts Xpeng (NYSE: XPEV), and Li Auto (NASDAQ: LI) have already launched smart driving systems based on the technology, AutoPix's report noted.

Nio conducted a review of the results of the end-to-end large model in November and found a number of issues, and is still looking for causes, according to the report.

The company's smart driving team is expediting the issues and plans to be ready to push out the new system in January, AutoPix said.

Nio's assisted driving feature that can allow a vehicle to drive autonomously from one parking space to another following navigation may not be rolled out with Banyan 3.1.0, the report said, adding that only Li Auto has rolled out the feature among domestic automakers so far.

Li Auto said on December 1 that it released OTA update version 6.5 in November, which leverages its end-to-end and visual language modeling (VLM) technologies to provide one-click point-to-point autonomous driving feature to all Li AD Max users.

The development of Nio's point-to-point autonomous driving capabilities is more complex than its peers.

The company is the only one among the mainstream car-making newcomers to adopt the battery swap model, and it has to realize that vehicles can perform autonomous battery swaps in urban areas based on the same set of technologies, in addition to the traditional parking space-to-park assisted driving function, the report said.

In the smart driving space, Nio is currently lagging behind its major peers in terms of marketing buzz, but industry sources say the actual gap isn't that big, AutoPix said.

The current end-to-end solutions of car companies actually consist of two or even more small AI models, such as Huawei's and Xpeng's, which have small models in charge of the sensing and regulation and control segments, the report said, citing a smart driving industry source.

Technically, this is still a segmented smart driving solution, according to the source.

Like its peers, Nio's end-to-end large model architecture, which will be launched early next year, is not technically one AI model to control it all, according to AutoPix.

Nio's NWM (Nio World Model), which was announced at its 2024 Tech Day event in July, can be thought of as a planning and control model with a larger number of parameters, AutoPix said.

Tesla (NASDAQ: TSLA) brought good results with the release of FSD V12 in early 2024. This has led to the end-to-end large model technology used by it becoming an industry consensus, and more car companies in China are beginning to experiment with this route.

Under the previous commonly adopted technology architecture, for an intelligent driving system, modules such as perception, prediction, decision-making, and control all require engineers in different fields to handle them.

The end-to-end intelligent driving system uses sensor data as input and is used directly in the vehicle's control commands, with the intermediate process being accomplished by neural network models.

Local media outlet LatePost reported in a June story that Nio's smart driving R&D department completed a structural change to focus more on end-to-end technology.

On July 11, Nio began rolling out the Banyan 2.6.5 CN system for vehicles based on the NT 2.0 technology platform, bringing AEB (automatic emergency braking) features optimized with end-to-end AI technology.

Nio ramps up car-buying promotions as it eyes higher sales in Dec