Alibaba Unveils Qwen 2.5 AI Model, Claims It Surpasses DeepSeek-V3

• Alibaba has released Qwen 2.5-Max, an AI model it claims surpasses DeepSeek-V3, GPT-4o and Llama-3.1-405B.
• The launch coincided with the Lunar New Year as Chinese technology companies responded to DeepSeek’s rapid rise and disruptive pricing.

Qwen 2.5-Max and the DeepSeek-V3 claim

Chinese tech giant Alibaba (9988.HK) has launched Qwen 2.5-Max, the latest version of its artificial intelligence model, claiming it outperforms the highly acclaimed DeepSeek-V3.

The Qwen 2.5-Max release coincided with the first day of the Lunar New Year. Alibaba’s cloud unit announced the model on its official WeChat account, referencing OpenAI and Meta’s most advanced AI models.

"Qwen 2.5-Max outperforms ... almost across the board GPT-4o, DeepSeek-V3 and Llama-3.1-405B,"

DeepSeek’s rapid rise and China’s AI competition

DeepSeek launched its DeepSeek-V3 model on January 10, followed by the R1 model on January 20. The startup’s disruptive pricing and efficiency have put pressure on U.S. tech firms, leading to a drop in tech stock prices and raising questions about the high development costs of AI models in the West.

DeepSeek’s rise has also intensified competition among Chinese technology companies. On January 22, ByteDance updated its flagship AI model, claiming it surpassed OpenAI’s o1 on AIME, a benchmark for complex instruction comprehension. The update followed DeepSeek’s assertion that its R1 model rivaled OpenAI’s o1 in performance.