For a while, the Chinese e-commerce giant Alibaba, who has been active in artificial intelligence, announced the new artificial intelligence model called “QWEN2.5-MAX”. The new model, built on QWEN2.5, reveals that even Alibaba’s attempted Chinese artificial intelligence attempted to be uneasy of Deepseek. Because Alibaba made this announcement in a period of holiday period in China.

According to Alibaba’s announcement via WeChat, QWEN2.5-MAX performs more beautifully than Deepseek-V3, GPT 4O and Llama-3.1-405B models. In order to make this statement, the company has put the new artificial intelligence model in various tests and the results obtained show that the structure is the truth.

Here are the test results published for QWEN2.5-Max

When we look at the tests of QWEN2.5-Max, we see that the most impressive result was taken in the Arena-Hard test. In the Livebench test, the artificial intelligence model, which left all its competitors behind, came in the third in the MMLU-PRRO facility and second in GPQA-Diamond and Livecodebench tests. It should be noted that the Arena-Hard test, where artificial intelligence is the first, is aimed at assumption of human preferences. In other words, QWEN2.5-Max became the artificial intelligence model that could think mostly compared to its competitors.

Alibaba has opened the new artificial intelligence model through the Owen Chat interface, which you can access through the contact. What the new model will offer in real use will be revealed with tests to be made by users.