Alibaba Launches Qwen3.8-Max, the Most Advanced AI Model Currently

Qwen 3.8 Max. (doc. alibaba)

JAKARTA - Alibaba officially launched Qwen3.8-Max, the latest and most advanced artificial intelligence (AI) model in the current Qwen family, which comes with 2.4 trillion parameters and is capable of processing up to 1 million tokens in one session.

The tech giant claims that this model ranks fifth in the Text Arena and second in the Vision Arena, with superior capabilities in coding, research, various work needs, and completing complex tasks.

Qwen3.8-Max is already available to developers worldwide via an API on Alibaba Cloud Model Studio. Meanwhile, the weights model that allows developers to run the AI on their own infrastructure will be released next week.

Qwen3.8-Max's capabilities in the field of coding are proven in internal testing, where this model successfully completed a software engineering project for 16 days by building an AI agent framework that can continue to evolve from scratch.

The trial reflects that Alibaba's most advanced AI model is capable of independently developing code and carrying out complex tasks without human intervention over long periods of time.

In addition to programming, this model can also be used for various professional needs, ranging from reviewing legal documents, financial research, sports analysis, application development, to modeling three-dimensional (3D) architecture.

Furthermore, Qwen3.8-Max is also able to analyze images and videos, which can process hundreds of pages of documents, television series, up to 100 hours of live broadcasts and turn them into a knowledge base that users can search.

The visual capabilities also allow Qwen3.8-Max to turn videos into vlogs, create animations from text commands, rebuild app interfaces from screenshots, and turn 2D floor plans into 3D interior visualizations.