Chinese tech giant Alibaba has once again accelerated the AI race, unveiling its largest language model yet, Qwen3-Max-Preview. Released on Friday, the new trillion-parameter model is positioned to compete directly with elite systems from OpenAI, Google, and Anthropic. In a notable strategy shift, Alibaba is keeping its most powerful model closed-source. A Trillion-Parameter Contender Enters the RingUnveiled on September 5, Qwen3-Max-Preview is Alibaba’s first model to cross the trillion-parameter threshold. In recent months, the company has unveiled a new flagship reasoning model, a powerful agentic coding model, and advanced video and image generation systems.