Nevertheless, another player is claiming the top position for AI. Alibaba, one of the world’s largest e-commerce companies, has announced the launch of its latest AI Reasoning Model, known as Qaven 3.
Alibaba claims that these models match the match (and even outperform) models in region works from both Google and Openai.
Qwen is not new in itself (as 3 names in the name reveal the name), but this is the first time Alibaba has actually tried to challenge top dogs, as well as its first step in more complex arguments.
Most models will soon be available for download from AI Dev companies under an open license, embracing the face or jithab. Their parameters will have some models, going from 0.6 billion to 235 billion.
Introduction Qwen3! We release and open-weight Qwen3, our latest large language models, including 2 Mo models and 6 dense models, ranging from 0.6B to 235B. Our flagship model, Qwen3-235B-A22B, coding, mathematics, general … pic.twitter.com/jwzkjehwcc get competitive results in benchmark assessment of our28 April, 2025
Parameters broadly translate the ability of a model to solve the problem in this example. More parameters, its capacity is equally better, but also costs longer and more.
Although he has not announced any more information, it is likely that these models will eventually make their way on Qwen Chatbot.
Models provide two modes. In thinking, the model will take time, step by step to respond with the best answer. In non-thinking mode, the model provides quick reactions for simple questions where motion is more important than depth.
“This flexibility allows users to control how much ‘thinking’ based on hand tasks performs. For example, difficult problems can be dealt with extended arguments, while easy people can be responded without direct delay,” Qwen team in a blog post,
“Intentedly, the integration of these two modes increases the ability of the model to implement stable and efficient thinking budget control.”
Along with Deepsek, Alibaba is an indication that China is catching with American AI companies. Deepsek has already performed at the same level, if not better, then compared to the choice of Google on chat and Google’s choice.
If Alibaba’s claims are correct, it is still another set of logic models that can match the performance of chat. Interestingly, it will also be a more powerful, smart model made by Deepsek.
Like the Deepsek, Alibaba seems to reduce the American market, offering AI technology which is only powerful but cheaper. However, it raises questions of morality and safety as many AI companies have been accused of cutting corners.
What is important about these models?
AI companies are going into a slightly new system for AI models. Originally, there was only one version of the AI ​​system, now there are many. These various systems are designed to handle various tasks, to try more for complex tasks and use low energy for easy challenges.
For example, Chatgpt has a chat 4o, which is its common model for most tasks, but for simple requests everyday, 4O is also mini. At her top, it has a series of logic models.
The Reasoning model is designed to take complex tasks. This is everything from coding to multi-step requests (asking it to do multiple things in a row) and deep research projects.
With this latest update from Quven, Alibaba is also aiming to do so, offering a model that can do simple everyday tasks, but also achieve the complex thinking that is being expected from AI today.
More than Tom’s guide
Back to laptop
