Alibaba Previews Qwen3.8-Max, a 2.4 Trillion-Parameter Multimodal Model

On July 19, 2026, Alibaba's Qwen team previewed Qwen3.8-Max-Preview, a 2.4 trillion-parameter model positioned as the next flagship in their Qwen family. The research team stated it is benchmarked as second only to Fable 5 among evaluated systems. This preview is currently accessible, though the full benchmark data, model card, and license are not yet released. The announcement coincided with the World AI Conference (WAIC) in Shanghai, occurring just two days after Moonshot AI launched Kimi K3, an open-weight model with 2.8 trillion parameters.

Alibaba confirmed that Qwen3.8 is slated for an open-weight release soon and described it as one of the most powerful models available, comparable to leading frontier systems. The preview build is available for purchase through Alibaba's Token Plan subscription service, offered at 10% of its standard pricing. Shuai Bai, a developer on the Qwen team, provided technical details, noting that Qwen3.8 is their first multimodal model exceeding 1 trillion parameters. This model is designed to handle text, images, video, and documents.

The company stated that Qwen3.8 is expected to outperform its predecessor, Qwen3.7-Max, in areas such as coding, full-stack development, data analysis, and office workflows. The preview release aims to provide early access to these advanced capabilities. The competitive landscape is intensifying, with major players like Alibaba and Moonshot AI rapidly advancing model scale and multimodal functionalities.