Alibaba’s Qwen3.8-Max AI model intensifies global AI race

Alibaba launched Qwen3.8-Max, a 2.4T parameter AI model with open weights coming soon, outperforming rivals in benchmarks and boosting shares, intensifying competition with OpenAI and Anthropic.

Alibaba has introduced Qwen3.8-Max, its most advanced AI model to date, boasting 2.4 trillion parameters and robust multimodal capabilities for both text and image processing. In internal benchmarks, Qwen3.8-Max reportedly outperforms Moonshot’s Kimi K3 and ranks just behind Anthropic’s Claude Fable 5, highlighting its competitive edge. The model is now available through Alibaba Cloud’s Model Studio APIs and business platforms, with early access discounts reaching up to 90%. Alibaba plans to release the model’s open weights next week, signaling a renewed commitment to open-source AI and enabling developers to download and customize the model for their needs. This strategic move intensifies competition with OpenAI and Anthropic, as Alibaba’s open-weight approach contrasts with their proprietary systems. The launch has boosted Alibaba’s shares by up to 7.3% and increased market confidence in its AI strategy, though prediction markets still favor Anthropic. Real-world enterprise performance of Qwen3.8-Max remains to be seen, as its architecture activates only a portion of its parameters per query.