Chinese AI Race Heats Up for New Year

China’s artificial intelligence sector is entering a new phase of competition as major tech firms unveil new models during the Lunar New Year period.

A year after DeepSeek disrupted the global industry with its powerful R1 and V3 models, rivals are accelerating development to avoid being overshadowed again.

DeepSeek is expected to introduce its next-generation V4 model soon. The company has already expanded its chatbot’s memory capacity dramatically, allowing it to process far larger volumes of information in a single task.

Other companies are moving quickly. ByteDance launched its Doubao 2.0 chatbot and Seedance 2.0 video-generation system, both designed for complex multi-step tasks. Alibaba is preparing its Qwen 3.5 model, while Zhipu introduced the open-source GLM-5 with stronger coding abilities.

Tencent has released a compact AI model optimized for consumer devices, and iFlytek unveiled Spark X2, trained entirely using domestic chips. NetEase Youdao and Dexmal also launched agent-based systems focused on automation and robotics.

Many of these tools aim to support the emerging “agent era”, where AI systems perform real-world tasks rather than simply generating responses.

The rapid wave of releases highlights China’s push to build competitive, lower-cost AI alternatives and reduce reliance on foreign technology.