%20--%3e%3cdefs%3e%3cstyle%3e%20.st0%20{%20fill:%20%23061b40;%20}%20.st1%20{%20fill:%20%23306af1;%20}%20.st2%20{%20fill:%20%235ce5cf;%20}%20%3c/style%3e%3c/defs%3e%3cg%3e%3cpath%20class='st0'%20d='M55,10.5h9v3h-9v9h12V7.5h-12v3ZM64,19.5h-6v-3h6v3Z'/%3e%3cpolygon%20class='st0'%20points='69%2016.5%2078%2016.5%2078%2019.5%2069%2019.5%2069%2022.5%2081%2022.5%2081%2013.5%2072%2013.5%2072%2010.5%2081%2010.5%2081%207.5%2069%207.5%2069%2016.5'/%3e%3cpolygon%20class='st0'%20points='95%2010.5%2095%207.5%2083%207.5%2083%2022.5%2095%2022.5%2095%2019.5%2086%2019.5%2086%2016.5%2095%2016.5%2095%2013.5%2086%2013.5%2086%2010.5%2095%2010.5'/%3e%3cpath%20class='st0'%20d='M40,1.5v21h11.6l1.4-1.4v-7.6h0c0,0-1.4-1.5-1.4-1.5l1.4-1.4V2.9l-1.4-1.4h-11.6ZM50,19.5h-7v-6h7v6ZM50,10.5h-7v-6h7v6Z'/%3e%3c/g%3e%3cpath%20class='st1'%20d='M23.1,24L14.7,4.8l-4.9,11.2h3.8l-1.8,4H3.3L12.1,0H2C.9,0,0,.9,0,2v20c0,1.1.9,2,2,2h21.1Z'/%3e%3cpath%20class='st2'%20d='M34,0h-16.8l10.6,24h6.2c1.1,0,2-.9,2-2V2C36,.9,35.1,0,34,0ZM32.5,20h-4V4h4v16Z'/%3e%3c/svg%3e)
In the fierce competition in artificial intelligence, Alibaba Cloud has once again made a move, officially launching its latest AI language model - Qwen3-Max. This model is not only the largest and most powerful product from the Tongyi team to date, with a total of over 1 trillion parameters, significantly enhancing multiple technical capabilities and bringing new shocks to the industry.
The Qwen3-Max-Instruct version has made a qualitative leap in code capabilities and agent performance, especially in areas such as knowledge reasoning, programming ability, instruction following, and multilingual understanding, where Qwen3-Max's performance is at the forefront of the industry. The Tongyi team at Alibaba Cloud also revealed that another version currently being trained, Qwen3-Max-Thinking, has shown extraordinary potential and is expected to be released to the public soon. This "thinking" version achieved a 100% accuracy rate in some high-difficulty reasoning tests, demonstrating its excellent reasoning ability.
In terms of training, the Qwen3-Max model used 36 trillion tokens for pre-training. Thanks to its unique MoE model structure design, the training process was stable and efficient. With the optimization of PAI-FlashMoE, the training efficiency of Qwen3-Max has been significantly improved, enabling support for up to 1M context training. At the same time, Alibaba Cloud has reduced time losses caused by hardware failures through various technical means, ensuring efficient training.
In practical applications, Qwen3-Max-Instruct has consistently ranked in the top three globally on the text leaderboard of LMArena, surpassing the well-known GPT-5-Chat. In the SWE-Bench Verified benchmark test focusing on real-world programming challenges, Qwen3-Max-Instruct achieved a high score of 69.6, demonstrating its outstanding programming capabilities. In the Tau2-Bench test measuring agent tool calling abilities, Qwen3-Max-Instruct also scored an excellent 74.8, successfully surpassing Claude Opus4 and DeepSeek-V3.1.
It can be said that Alibaba Cloud's Qwen3-Max not only leads in technical parameters but also demonstrates strong capabilities in application performance, injecting new momentum into the future development of artificial intelligence.