Open-Source Breakthrough Under Chip Blockade: Alibaba Launches 'Qwen3.8-Max' to Reshape Global AI Landscape with Low-Cost Strategy

Chinese tech giant Alibaba launched its most advanced artificial intelligence model, 'Qwen3.8-Max,' on Monday (August 3), marking a significant milestone in the global AI competition. With 2.4 trillion parameters, the model supports multimodal processing of text, images, and video, and can handle up to one million tokens in a single input—surpassing many Western counterparts in context length.

The model quickly rose to the top of China-based entries on Arena.AI, a leading AI benchmarking platform, and contributed to a 7% surge in Alibaba's Hong Kong-listed shares. Unlike traditional dense models, Qwen3.8-Max uses a Mixture-of-Experts (MoE) architecture, activating only about 95 billion of its 2.4 trillion parameters during inference. This design drastically reduces computational costs and latency, making it highly efficient for enterprise deployment.

Qwen3.8-Max is set to be officially released next week via Alibaba Cloud's Model Studio platform. The company claims the model completed a full software engineering project within 16 days, showcasing its practical utility.

Meanwhile, Chinese AI startup DeepSeek continues its aggressive low-cost strategy with the release of V4-Flash on July 31. According to Artificial Analysis, V4-Flash costs just $0.14 per million input tokens and $0.28 per million output tokens—significantly cheaper than Claude Fable 5 ($3.15) and GPT-5.6 Sol ($1.86). The average cost per benchmark task is estimated at $0.03, compared to $0.86 for Moonshot AI's Kimi K3.

Both Alibaba and DeepSeek are embracing 'open-weight' models, allowing developers to download and fine-tune core model parameters. This contrasts with the closed models of OpenAI, Anthropic, and Google. Omdia's chief analyst Lian Jye Su noted that many enterprises prioritize affordability, transparency, and accessibility over cutting-edge performance—needs that open-weight models fulfill.

This strategic shift highlights China's growing influence in AI, leveraging software innovation to counter U.S. semiconductor export restrictions. As open-weight models gain global traction, they could reshape enterprise AI adoption, particularly in cost-sensitive markets. Alibaba's move signals a broader push to dominate the B2B AI infrastructure space through its cloud ecosystem.