According to industry sources on April 22, Tencent and Alibaba are seeking to participate in the latest funding round of DeepSeek, a rising star in domestic large models. Driven by the competition from these tech giants, DeepSeek's valuation target has surged from $10 billion a few days ago to over $20 billion.

Previously on April 18, DeepSeek had reported seeking to raise $300 million at a valuation of no less than $10 billion. The sharp increase in its capital appeal reflects the market's premium expectations for leading independent model vendors. On the product side, DeepSeek is expected to launch its new generation model, DeepSeek v4, within this month. The core upgrade of this model lies in the introduction of multimodal processing capabilities, aiming to further expand the boundaries of industry applications.

DeepSeek

Meanwhile, Tencent's self-developed model, Hunyuan, has also made significant progress. It is expected to officially launch Hunyuan 3.0 this week. According to reports, this version is led by Yao Shunyu, a former core researcher at OpenAI, and may adopt a lightweight design with approximately 30B parameters, aiming to maintain high performance while reducing deployment costs and inference latency.

The current large model market is showing a clear polarization trend: on one hand, leading startups are experiencing rapid valuation growth, with capital concentrating heavily on superior algorithm teams; on the other hand, tech giants are pursuing extreme "smallness" and efficiency balance in model architecture. With DeepSeek's completion of multimodal capabilities and Hunyuan 3.0's lightweight entry, the competition among domestic models has shifted from a simple parameter scale race to a comprehensive contest of multimodal capabilities and potential for edge-side deployment. This could reshape the investment and financing landscape and technical standards of the domestic AI industry.