Alibaba Releases the Qwen 2.5 AI Model and Says It Beats DeepSeek-V3

Alibaba has increased rivalry in China’s AI market by releasing Qwen 2.5, an improved AI model that it says performs better than DeepSeek-V3.

Alibaba, a Chinese internet behemoth, has released Qwen 2.5-Max, a new version of their AI model that it claims outperforms the much-lauded DeepSeek-V3.

Alibaba’s cloud division stated in a statement shared on its official WeChat account that “Qwen 2.5-Max outperforms … almost across the board GPT-4o, DeepSeek-V3, and Llama-3.1-405B,” alluding to the most cutting-edge open-source AI models from OpenAI and Meta.

The news highlights the need to remain competitive in the face of DeepSeek’s explosive growth and was made on the first day of the Lunar New Year, when most Chinese companies take a break for festivities.

Both Silicon Valley and local competitors have been rocked by the rise of the Chinese AI firm DeepSeek.

Since its January introduction, the DeepSeek-V3 model has garnered international attention due to its great efficiency and low operating expenses. Days later, the business unveiled DeepSeek-R1, which it says competes with OpenAI’s most recent models. Concerns over the high expenditures by top AI companies in the US have been raised by this quick invention, which has not only sparked a race among Chinese tech companies but also led global tech stocks to plummet.

Alibaba’s cloud business has responded by stepping up its AI skills, while rivals like ByteDance, the parent firm of TikTok, have also improved their AI models to meet DeepSeek’s standards. Two of the largest tech companies in China, Baidu and Tencent, are also closely following suit.

In a rare July interview with Chinese media outlet Waves, Liang Wenfeng, the CEO of DeepSeek, stated that the company “did not care” about pricing wars and that its primary objective was to achieve artificial general intelligence, or AGI.

AGI is defined by OpenAI as autonomous systems that outperform humans in the majority of economically significant tasks.

Liang contrasted DeepSeek’s lean operations and flexible management style with the exorbitant expenses and top-down structures of China’s major tech companies, saying in his July interview that he thought they would not be well suited to the future of the AI business.

“The capabilities of tech giants are limited, and large foundational models require ongoing innovation,” he stated.

Alibaba’s large workforce and conventional corporate structure stand in stark contrast to the startup’s lean structure, which is mostly staffed by young academics and PhDs from prestigious Chinese institutions.

Because DeepSeek’s prior model, DeepSeek-V2, offered token processing for under 1 yuan ($0.14) per million tokens, it had already sparked a pricing war in China’s AI market. Alibaba Cloud was had to cut its own AI prices by 97% as a result, and other companies were under pressure to do the same.

Notwithstanding the financial consequences, DeepSeek is more concerned with expanding the frontiers of AI innovation than with pricing competition.

Alibaba’s Qwen 2.5-Max launch marks a new chapter in China’s AI growth as the competition for dominance in AI heats up, with tech titans vying for market share with DeepSeek’s game-changing innovations.