This time it’s Alibaba’s Qwen
Alibaba Group Holding has unveiled its new flagship AI model, Qwen3.8-Max. The model delivered impressive results: experts are placing it alongside leading global developments, including Anthropic’s Fable 5, which was previously placed under temporary US export controls due to its advanced capabilities.
According to the company, Qwen3.8-Max handles autonomous coding and long-horizon tasks well: during internal testing, the model independently ran a software engineering project for 16 days. Developers also improved the system’s efficiency — it now activates only part of its parameters when processing a request, reducing compute costs and latency.
On a number of benchmarks, the new system outperforms the recent release of Kimi K3 from Moonshot AI. Chinese developers have shown they can rapidly close the gap with market leaders like Anthropic and OpenAI, despite limited access to computing resources. Another player, DeepSeek, expanded access to its latest model, V4 Flash, on July 31.
Ling Vey-Sern, managing director at Union Bancaire Privee, said the market continues to underestimate Chinese AI models due to US chip restrictions and general skepticism. In his view, the real gap with leaders is much smaller and continues to narrow quickly, with the Qwen3.8 release serving as further proof of this trend, following Kimi K3.
Following the announcement, Alibaba’s shares rose by nearly 7.3% in Hong Kong on Monday. It’s the biggest gain in almost a month.
