Yuanrong Qixing CEO Zhou Guang: To solve the "seesaw effect" of small models, we must shift to the paradigm of large models

2026-04-12 08:04:04

On April 12th, at the High-Level Forum on the Development of Intelligent Electric Vehicles, Zhou Guang, CEO of Yuanrong Qixing, mentioned that by 2025, the user stickiness of urban NOA (Navigation on Autopilot) will be approximately in the 20%-30% range, with the penetration rate of urban NOA around 15%. Zhou believes that the most fundamental reason behind this is: today’s mass-produced small models for intelligent driving have very few parameters, mostly below 1 billion, and even below 0.1 billion, with typical operational computing power of 100-200 TOPS. To solve the “scalability dilemma” of small models, it is necessary to shift from a small model paradigm to a large model paradigm, systematically enhancing capabilities through scaling.

View Original

This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.