Alibaba Launches Qwen3.8-Max with 2.4 Trillion Parameters
Alibaba released Qwen3.8-Max on August 3, 2026, its largest AI model to date, featuring 2.4 trillion total parameters in a mixture-of-experts architecture. The model is available via Alibaba Cloud APIs and is priced at $2 per million input tokens.
Alibaba launched its most powerful AI model on August 3, 2026. Qwen3.8-Max features 2.4 trillion total parameters in a mixture-of-experts architecture that activates only 95 billion parameters per token during inference.
The model supports a context window of up to 1 million tokens, which allows it to process hundreds of pages of documents or roughly 100 hours of video in a single request. It handles multimodal inputs including text, images, and video.
In internal tests, Alibaba said the model completed a full 16-day autonomous software engineering project from scratch and ranked competitively against models from Anthropic and Moonshot AI's Kimi K3 on coding and research benchmarks. Independent verification of those results has not yet been published.
The model is available immediately through Alibaba Cloud's Model Studio APIs. Pricing starts at $2 per million input tokens and $6 per million output tokens. Alibaba plans to release open weights for Qwen3.8-Max and a smaller 27-billion-parameter version around August 10.
Alibaba's US-listed shares rose about 4.5 percent and Hong Kong shares climbed roughly 7 percent following the announcement, reflecting investor confidence in the company's AI strategy.
The release intensifies competition between Chinese and US AI labs. Chinese models have increasingly matched or approached the performance of leading Western systems while often undercutting them on price, which is pressuring US companies on both cost and open-weight strategy.
