Alibaba Previews Qwen3.8-Max, a 2.4 Trillion-Parameter Multimodal Model, Days After Moonshot’s Kimi K3 Open-Weight Launch

Alibaba's artificial intelligence division has unveiled Qwen3.8-Max-Preview, a massive multimodal model boasting 2.4 trillion parameters, marking another significant milestone in the intensifying competition for AI dominance. The preview launch arrives just days after Moonshot AI released its Kimi K3 open-weight model, underscoring the rapid pace of innovation in large language models. Alibaba positions Qwen3.8-Max as a frontier-class system, claiming it ranks second only to OpenAI's o1 model in performance capabilities.

The Qwen3.8-Max-Preview is currently accessible through Alibaba's Token Plan, Coder, and CoderWork platforms at a promotional rate of 10% of standard pricing. However, the release remains notably preliminary, lacking several critical technical specifications. Alibaba has not yet published benchmark results, comprehensive model cards, explicit licensing information, standard per-token pricing structures, or detailed technical documentation. The model employs a Mixture-of-Experts (MoE) architecture, a design pattern that activates specific subsets of parameters for different inputs, enabling efficient scaling to massive parameter counts while maintaining computational efficiency.

• Accelerating competition: The rapid succession of major model releases demonstrates intensifying competition between Chinese tech giants and Western AI leaders
• Multimodal capabilities: The model's multimodal design suggests advanced reasoning across text, images, and potentially other data types
• Preview strategy: Releasing preview versions allows companies to gather user feedback and benchmark performance before formal publication
• Transparency concerns: The absence of published benchmarks raises questions about independent verification of performance claims
• Pricing accessibility: The 90% discount during preview phase could drive adoption among developers and researchers

Alibaba's Qwen3.8-Max-Preview represents a critical juncture in the global AI arms race, particularly for Chinese technology companies establishing competitive alternatives to Western models. The timing and scale of the release demonstrate that cutting-edge AI development extends beyond Silicon Valley, with major technology ecosystems developing comparable frontier models. As enterprises and developers evaluate AI solutions, the proliferation of advanced models increases options while simultaneously raising challenges around benchmark standardization, safety verification, and regulatory compliance across different regions and use cases.