DeepSeek seems to have exploded in popularity almost overnight. With the release of DeepSeek-R1 on January 20th, the new AI model has been making waves in the industry and has already had a massive impact on the stock market. While the AI company has been around for a little while (as far as AI companies go), its latest AI model has put it in the spotlight. DeepSeek primarily functions like ChatGPT or any other LLM AI model, but what sets it apart is its significantly lower cost. Reports suggest that DeepSeek was built for around $6 million, compared to some US-based models that cost billions.

And just as we were wondering how Western companies would respond to this DeepSeek boom and its cheaper approach to building LLMs, another AI model has entered the game – and it’s not from an overseas rival. On Wednesday, Chinese tech giant Alibaba released a new version of its Qwen 2.5 artificial intelligence model, claiming it surpasses not only the highly acclaimed DeepSeek-V3 but also nearly every major AI model on the market.

DeepSeek faces competition both globally and domestically

Qwen 2.5 arrives at a time when people in China are off work and celebrating with their families, as it launches on the first day of the Lunar New Year. This highlights just how much of an impact DeepSeek has had on the AI market, with competitors not only from the Western world but also within China looking to challenge this overnight sensation. An expert recently claimed that DeepSeek could bring “the end of closed-source AI” thanks to its open-source nature and lower hardware requirements.

In an announcement posted on Alibaba’s official WeChat account, the company's cloud unit compared its latest model to OpenAI and Meta’s most advanced open-source AI models, stating, “Qwen 2.5-Max outperforms … almost across the board GPT-4o, DeepSeek-V3, and Llama-3.1-405B.”

Interestingly, Alibaba isn't the only Chinese company stepping up against DeepSeek's rapid rise. Just two days after the release of DeepSeek-R1, TikTok owner ByteDance introduced an update to its flagship AI model, claiming it outperformed Microsoft-backed OpenAI’s GPT-4o in AIME—a benchmark test that evaluates how well AI models understand and respond to complex instructions.

Not the first time Alibaba has taken a shot at DeepSeek

Unlike companies like Nvidia, which have praised the new DeepSeek-R1 model as “an excellent AI advancement,” Alibaba has been going head-to-head with DeepSeek. The predecessor to DeepSeek-V3, DeepSeek-V2, triggered an AI model price war in China after its release last May. As an open-source model, DeepSeek-V2 was unprecedentedly cheap, costing just 1 yuan ($0.14) per 1 million tokens—the units of data processed by the AI model. This led Alibaba's cloud unit to announce price cuts of up to 97% across a range of models.

DeepSeek’s founder, Liang Wenfeng, responded to this price war saying he “did not care” and his primary goal is achieving AGI (artificial general intelligence). While tech giants like Alibaba have hundreds of thousands of employees, DeepSeek operates more like a research lab as Liang believes that “large foundational models require continued innovation, and tech giants' capabilities have their limits.”

We have yet to see more concrete examples of Alibaba's Qwen 2.5 outperforming DeepSeek and other AI language models, but it's hard not to recognize the impact DeepSeek has had in such a short time. With its low computational requirements and cost efficiency, it has already caused a significant shift in how companies will approach AI model development moving forward.

Source: Reuters

About the Author

Hassam Nasir

Hassam boasts over seven years of professional experience as a dedicated PC hardware reviewer and writer.