China’s Alibaba introduced its artificial intelligence model, the Qwen3.8-Max, on Monday. The Qwen3.8-Max is said to be Alibabas most capable artificial intelligence model. It is similar in size to a model launched by Moonshot AI last month.
Chinese tech companies are very important in the world of intelligence models. They are competing with each other to build powerful systems that are not too expensive to use.
The Qwen3.8-Max has 2.4 trillion parameters. Parameters are like settings that the model learns from data. The model uses these settings to find patterns, generate answers, and do tasks. Moonshots model, Kimi K3 has 2.8 trillion parameters.
Just because a model has a lot of parameters does not mean it is better.. Parameters are a good way to measure how powerful a model is.
Chinese tech companies like to share the number of parameters in their models. This helps their models become popular among developers. The models are also weight-trained, which means that developers can download and use them.
On the other hand, companies like OpenAI, Anthropic and Google do not share the number of parameters in their models.
The Qwen3.8-Max was introduced on a website called Arena.AI. This website compares artificial intelligence models. The Qwen3.8-Max became the ranked Chinese model for text tasks.. It is still behind some models from Anthropic.
However when it comes to analyzing images and videos the Qwen3.8-Max is the best model in the world. It is behind a model from Anthropic.
Both the Qwen3.8-Max and the Kimi K3 can handle text, images and videos. They can process up to 1 million tokens at a time. Tokens are like pieces of data.
Alibaba said that its model uses a design that divides tasks among different parts of the system. This means that 95 billion parameters are used at a time. This reduces costs. Makes the model faster.
Alibaba said that the Qwen3.8-Max completed a software project in 16 days. The Qwen3.8-Max will be available week, through Alibaba Clouds Model Studio platform.
