Alibaba has launched Qwen3.8-Max, a large language model designed to handle autonomous coding, research and complex real-world tasks with limited human supervision.
According to the company, the model contains 2.4 trillion parameters and supports a context window of up to 1 million tokens. This allows it to process large volumes of documents, instructions, research materials and software code within a single task.
Qwen3.8-Max can generate code, test and debug applications, preview results, analyse system logs and improve its output through repeated iterations. Alibaba said the model is built to manage longer engineering and research workflows rather than provide isolated coding suggestions.
As a demonstration, Alibaba used the model to build a self-evolving AI agent framework called oh-my-cli. The system uses user feedback, community practices and data from its own testing to refine its performance. Alibaba has released the framework as open source on GitHub.
The launch comes amid rising competition among Chinese AI developers. In July, Moonshot AI released Kimi K3, which it described as the largest open-source AI model by parameter count at approximately 2.8 trillion parameters. The race is increasingly focused on reasoning, coding and autonomous task execution alongside model size.
Source: EntArabi
