Alibaba Advances in AI Race with New Qwen3 Model

Alibaba Cloud has unveiled its Qwen3-Next architecture, marking a significant leap in open-source AI development. The new Qwen3-Next-80B-A3B model operates 10x faster than its April predecessor while slashing training costs by 90%, demonstrating Chinese tech firms' aggressive push to compete with US AI leaders.

The hybrid attention system and mixture-of-experts design enable superior text processing with consumer-grade hardware. A specialized reasoning variant, Qwen3-Next-80B-Thinking, reportedly outperforms Google's Gemini-2.5-Flash in benchmark tests, signaling Alibaba's growing technical prowess.

By open-sourcing through GitHub and Hugging Face, Alibaba positions Qwen as a cornerstone of global AI infrastructure. This strategic move coincides with reported Apple collaborations, suggesting expanding industry partnerships amid intensifying AI competition.