The landscape of artificial intelligence development for coding supremacy is narrow, characterized by intense competition and rapid advancements. Currently, the industry focuses on two major players, with Anthropic’s Claude Fable 5 acting as the dominant benchmark for performance, particularly for Chinese tech organizations.

Following a highly anticipated reveal, Alibaba Group Holding Limited saw its shares climb by 5.4% in Hong Kong trading on Monday, July 20. This surge followed the company’s preview of Qwen 3.8 Max, its latest and most powerful large language model designed for coding. Bloomberg reported that Alibaba positions Qwen 3.8 Max as trailing “only Fable 5 among today’s leading systems.”

This announcement comes shortly after another major Chinese AI lab, Moonshot AI, made a similarly ambitious claim. On July 17, Moonshot introduced Kimi K3, which also pitched itself as a near-equal competitor to the global standard.

Alibaba’s strategic announcement is complicated by its recent actions against Anthropic. On July 10, Alibaba prohibited its employees from using Anthropic’s Claude Code for professional work. This decision followed the discovery that Anthropic’s tool could detect whether users were operating from China, which raised significant data security and compliance concerns.

This prohibition adds friction to an already competitive situation. Recently, Anthropic accused operators linked to Alibaba’s Qwen lab of managing approximately 25,000 fraudulent accounts designed to extract outputs from Claude through distillation, an adversarial technique used to train competing models.

Despite these challenges, Alibaba is aggressively promoting its own alternative, Qoder, a coding platform that now hosts the Qwen 3.8 Max preview. This highlights a clear trend where Alibaba is replacing an external, rival system with an internally developed one that it claims is nearly as capable.

A larger pattern is emerging among China’s top AI companies. In June, Zhipu AI also positioned its GLM 5.2 model within striking distance of Opus 4.8 on a key benchmark, promising similar performance at a lower cost. Moonshot AI’s Kimi K3, which boasts 2.8 trillion parameters, made its identical pitch just days before Alibaba, a move that momentarily impacted global tech stocks and led Moonshot to inform investors of plans for an initial public offering within six months.

Crucially, none of these ranking claims are verified by independent benchmarks. Alibaba has not released a model card or standard benchmark tables for Qwen 3.8 Max, meaning its claim that the model is trailing “only Fable 5” is based solely on internal testing.

Alibaba’s primary differentiator, however, might be accessibility. The company has promised to release Qwen 3.8 as an open-weight model soon. This is a significant detail, as it directly contrasts with Anthropic’s models. The U.S. government temporarily restricted foreign access to Claude Fable 5 and Mythos in June before only partially restoring it in July. An open-weight model, which can be downloaded and run on a company’s own servers, cannot be disabled by an export control, making it an attractive asset in the current geopolitical climate.

Alibaba also holds a complex position as it owns a 36% stake in Moonshot AI. This gives the conglomerate financial exposure to two of its primary competitors in the domestic and open-weight markets. The market’s response, including the stock rally, reflects a scenario where Alibaba wins regardless of which company performs better, securing the company’s influence over the future of the AI coding landscape.