Alibaba (9988.HK) has unveiled its latest artificial intelligence model, Qwen 2.5, in a strategic move to reinforce its standing in China’s AI race.
Released on the first day of the Lunar New Year, a time when most of China is on holiday, the timing signals an urgent response to the rapid rise of local rival DeepSeek.
The new model, Qwen 2.5-Max, claims to outperform OpenAI’s GPT-4o, DeepSeek-V3, and Meta’s Llama-3.1-405B across multiple AI benchmarks.
The announcement marks Alibaba’s latest attempt to regain ground in the AI arms race as DeepSeek’s disruptive pricing and efficiency threaten to shift market dynamics.
The AI battle in China has intensified in recent weeks, with firms such as ByteDance and Baidu also scrambling to upgrade their models to stay competitive.
Unlike previous AI developments, where China trailed US companies, local firms are now setting global standards.
This shift has prompted not only competition among Chinese firms but also increased scrutiny of AI spending by major US tech players.
DeepSeek shakes up China’s AI market
DeepSeek’s rapid emergence has triggered a chain reaction in China’s AI sector, forcing established players such as Alibaba, Tencent, and Baidu to respond aggressively.
The startup’s AI assistant, launched on Jan. 10, followed by its R1 model on Jan. 20, has positioned it as a serious challenger to entrenched AI firms.
Investors have taken notice, with DeepSeek’s cost-efficient model prompting questions about the sustainability of high R&D expenditures at US-based AI firms.
DeepSeek’s rise mirrors previous disruptions in China’s tech landscape, where cost-cutting innovations have led to major industry shifts.
In May 2023, the launch of DeepSeek-V2 triggered a price war in AI services, forcing Alibaba Cloud to slash model access fees by up to 97%.
Now, with the release of DeepSeek-V3, the stakes have risen further, compelling Alibaba to fast-track its own AI advancements.
Alibaba’s latest AI model builds on its Qwen series, which has been central to its strategy of competing in both domestic and international AI markets.
While Alibaba’s cloud division has struggled against DeepSeek’s momentum, the Qwen 2.5 update suggests a recalibrated approach, focusing on performance benchmarks rather than just pricing competition.
AI price war escalates
The Chinese AI market is experiencing a pricing war that may significantly impact the long-term profitability of tech giants.
With DeepSeek’s aggressive pricing—reportedly offering AI processing at just 1 yuan ($0.14) per million tokens—competitors have been forced to reconsider their business models.
Alibaba, Baidu, and Tencent have already introduced steep price cuts on AI services, mirroring the pattern seen in China’s cloud computing and e-commerce industries.
Despite cost concerns, companies remain locked in a battle for dominance.
ByteDance, the owner of TikTok, quickly followed DeepSeek’s R1 release with an update to its own AI model, claiming it outperformed OpenAI’s o1 on key performance indicators.
This suggests that the AI race is no longer just about catching up with the US but about establishing China as a leader in AI innovation.
With government backing and an increasingly competitive ecosystem, China’s AI sector is positioning itself as a global force.
While Alibaba’s Qwen 2.5 may not be the final word in this battle, its release underscores the urgency among Chinese tech giants to stay ahead of emerging disruptors like DeepSeek.
The real challenge will be balancing cost efficiency with sustainable AI development, as companies navigate a market where margins are shrinking but competition is escalating.
The post Alibaba unveils Qwen 2.5 AI model, says it outperforms GPT-4o and DeepSeek-V3 appeared first on Invezz