One year after Chinese start-up DeepSeek unsettled global markets with a low-cost artificial intelligence model, domestic rivals are preparing a new wave of product launches aimed at strengthening their position in China’s rapidly evolving AI sector.
The Hangzhou-based company’s rise during the 2025 Spring Festival holiday reshaped China’s AI ecosystem, pushing affordable, open-source models to the forefront despite U.S. export controls limiting access to advanced semiconductors.
With this year’s Spring Festival beginning on February 15, several Chinese firms are expected to unveil upgraded models, signalling a more competitive and mature domestic market.
High Expectations After DeepSeek Breakthrough
DeepSeek’s debut stunned the industry by demonstrating that a powerful AI model could be developed at a fraction of the cost of Western counterparts.
“The surprise would be if some of these new models end up being underwhelming. I think there are high expectations here,” said Alfredo Montufar-Helu, managing director at Ankura Consulting in Beijing.
Zhipu AI on Wednesday released a new model featuring enhanced coding capabilities and the ability to carry out extended tasks without further user prompts.
ByteDance followed with the official unveiling of Seedance 2.0, described by the state-backed Global Times as capable of generating cinematic-quality video content within seconds. The company is also expected to upgrade its Doubao chatbot, currently China’s most widely used AI app with more than 155 million weekly active users.
DeepSeek is preparing its next-generation V4 model, while Alibaba is widely expected to introduce its Qwen 3.5 series, offering stronger mathematical reasoning and coding capabilities. Support code submissions to open-source platform Hugging Face suggest an imminent release.
None of the companies have announced formal launch dates.
Low-Cost, Open-Source Model Becomes the Norm
DeepSeek’s initial release in January 2025 triggered a global technology sell-off, wiping nearly $600 billion from Nvidia’s market value in a single session and prompting competitors to rapidly upgrade their own systems.
In recent years, DeepSeek has consistently undercut rivals on pricing, driving usage costs well below many U.S. offerings. A report by research group RAND found Chinese AI systems operate at roughly one-sixth to one-quarter of the cost of comparable American models.
“DeepSeek showed the industry that you can create a very good model even when you’re resource-constrained,” said Lian Jye Su, chief analyst at Omdia.
Open-source access has since become a defining feature of China’s AI strategy. Companies including Baidu, Tencent and ByteDance have expanded public access to their models via platforms such as Hugging Face.
From Core Models to Consumer Integration
While DeepSeek remains focused on advancing foundational model performance, rivals are increasingly shifting towards consumer-facing applications.
Alibaba’s Qwen chatbot, for instance, has experimented with enabling purchases directly through conversational prompts, reflecting pressure to monetise AI investments.
DeepSeek remains structurally distinct. Its parent company is a quantitative hedge fund controlled by founder Liang Wenfeng, allowing it to prioritise research over rapid commercialisation and shareholder returns.
As China’s AI sector enters its second year of post-DeepSeek competition, expectations remain high that the next generation of models will further narrow the gap with global leaders or redefine the rules altogether.
(with inputs from Reuters)
