Alibaba has announced the release of the Qwen-3-Max-Preview, its most advanced large language model so far. The system, launched on September 8, 2025, contains more than 1 trillion parameters and is now accessible on Alibaba Cloud and OpenRouter.
The debut highlights Alibaba’s ambitions in the rapidly intensifying AI competition, putting it in direct comparison with rivals such as OpenAI’s GPT-4.5, believed to operate with between 5 and 7 trillion parameters, and models from Google DeepMind.
This version represents a leap forward from Alibaba’s July rollout of Qwen3-235B-A22B-2507, offering far greater scale and capability. Aimed at enterprises, developers, and researchers, the model comes with a premium pricing structure: $0.861 per million input tokens and $3.441 per million output tokens. These costs place it among the higher-priced products in the Qwen family, reflecting its positioning for advanced use cases rather than casual adoption.
By launching Qwen-3-Max-Preview, Alibaba signals that it intends to compete not only in China but also on the global AI stage. Analysts suggest that the company’s decision to focus on tiered pricing and enterprise-grade performance indicates a strategy aimed at businesses requiring high-volume processing and cutting-edge capabilities.
The release reinforces the trend of AI model scaling as a benchmark of innovation, with technology giants worldwide racing to deliver ever-larger systems that balance raw power with accessibility. For Alibaba, the Qwen-3-Max-Preview represents both a technological milestone and a statement of intent in the ongoing AI race.
