Alibaba releases Qwen 2.5-Max AI model, claims superiority over DeepSeek-V3 and rivals

Alibaba unveiled its latest AI model, Qwen 2.5-Max, yesterday, asserting it outperforms leading models like OpenAI’s GPT-4, Meta’s Llama-3.1-405B, and the headline-grabbing DeepSeek-V3.

The release coincided with the Lunar New Year holiday, highlighting the competitive pressure from DeepSeek’s rapid rise in China’s AI sector.

Alibaba’s cloud unit stated in a WeChat announcement that Qwen 2.5-Max excels “almost across the board” against its rivals. The timing suggests an urgency for Chinese AI players to keep pace with DeepSeek, which has disrupted the industry with its low-cost, high-performance models.

DeepSeek’s recent releases—the DeepSeek-V3-powered AI assistant on January 10 and the R1 model on January 20—have shaken global tech markets, prompting investors to question the high spending of US AI firms. The startup’s success has also spurred Chinese competitors, including ByteDance, to accelerate their own AI advancements.

DeepSeek’s predecessor, DeepSeek-V2, ignited a price war in China last year by offering open-source AI capabilities at just 1 yuan (US$0.14) per 1 million tokens. This forced Alibaba, Baidu, and Tencent to slash prices by up to 97% on their AI services.

Liang Wenfeng, DeepSeek’s founder, emphasized in a July interview that the company prioritizes achieving artificial general intelligence (AGI) over price competition. He contrasted DeepSeek’s lean, research-focused approach with the high costs and rigid structures of China’s tech giants, suggesting they may struggle to lead in AI innovation.

Alibaba’s latest release signals intensifying competition in China’s AI market, with implications for cost efficiency and innovation in the global AI race.

Share this Post:

Accessibility Toolbar