BEIJING, Feb 12 (GeokHub) — One year after DeepSeek’s breakthrough low-cost artificial intelligence model shook global tech markets, Chinese AI companies are ramping up efforts to release upgraded models, competing to capture both consumer and enterprise attention.
DeepSeek, the Hangzhou-based startup whose first-generation model debuted during China’s Spring Festival in early 2025, disrupted the AI landscape by offering high-performance, open-source models at a fraction of the cost of comparable U.S. systems. Its launch not only shook investor confidence in established players but also established low-cost, open-source AI as a new industry standard.
Chinese Firms Gear Up for Next-Generation AI
This year, DeepSeek will face competition from domestic rivals during China’s Spring Festival holiday, beginning February 15. ByteDance unveiled its Seedance 2.0 video-generating AI, capable of producing cinematic-quality content with minimal prompts. ByteDance is also preparing upgrades for Doubao, China’s most popular AI chatbot, boasting over 155 million weekly active users.
Zhipu AI recently released a model with advanced coding capabilities and the ability to run long-duration tasks autonomously. Alibaba is expected to launch its Qwen 3.5 series soon, featuring enhanced reasoning and coding functions, while DeepSeek is preparing its V4 model, focusing on core AI performance.
Industry experts note that these upgrades are highly anticipated. “The surprise would be if these new models are underwhelming. Expectations are very high,” said Alfredo Montufar-Helu, managing director at Ankura Consulting in Beijing.
Low-Cost, Open Source Becomes the Norm
DeepSeek’s 2025 launch reshaped global perceptions of AI production, proving that world-class models could be developed efficiently under resource constraints. Its pricing strategy — offering models at roughly one-sixth to one-fourth the cost of U.S. alternatives — forced Chinese competitors to adopt similar low-cost, open-source approaches.
Lian Jye Su, chief analyst at Omdia, said, “DeepSeek demonstrated that strong reasoning capabilities, open-source access, and low deployment costs could redefine how foundation models are built and deployed.”
Before DeepSeek, some Chinese tech leaders, including Baidu CEO Robin Li, had argued that closed-source systems would dominate. Following DeepSeek’s success, however, major companies such as Baidu, ByteDance, and Tencent opened portions of their AI models to developers, contributing to a surge of innovation on open-source platforms like Hugging Face.
Consumer-Centric AI Gains Focus
While DeepSeek prioritizes research over commercialization, its rivals are increasingly integrating AI into consumer services. Alibaba’s Qwen chatbot has been experimenting with enabling users to purchase goods directly through conversational prompts, reflecting shareholder and market pressure to monetize AI investments.
In contrast, DeepSeek’s parent, a quantitative hedge fund controlled by founder Liang Wenfeng, allows the company to focus on advancing model performance without external investor constraints. This structural distinction has positioned DeepSeek as a research-first competitor amid a rapidly commercializing AI landscape.
The flurry of model releases in 2026 signals China’s continued ambition to lead in low-cost, high-performance AI while blending consumer applications with foundational AI research.
