Alibaba Group officially launched its latest flagship AI model, Qwen3.8-Max, on August 3, 2026 [1, 2, 3, 4, 5, 6]. The model boasts 2.4 trillion parameters, making it one of the largest AI models developed by the company to date [1, 3, 4, 5, 6]. Alibaba described Qwen3.8-Max as capable of handling complex tasks autonomously, excelling in areas such as deep research, programming, autonomous coding, long-horizon execution, and image analysis [1, 3, 4, 5, 6].
Alibaba claims Qwen3.8-Max performs at a level rivaling leading global models, including Anthropic's Fable 5, and sometimes surpasses it on benchmark tests [1, 3, 4, 6]. However, some sources indicate the model does not fully reach "second only to Fable 5" status in benchmark testing [2]. Despite this, Qwen3.8-Max ranks highest among Chinese AI models on text-based tasks at Arena.AI, though some Anthropic models remain ahead overall [5, 6].
The model outperforms Chinese competitor Moonshot AI’s Kimi K3, which has 2.8 trillion parameters, on several benchmarks despite having fewer parameters [1, 3, 4, 5, 6]. To improve efficiency, Qwen3.8-Max activates only parts of its parameter set during use, reducing computational cost and latency [1, 3, 4, 6].
Ling Vey-Sern, managing director at Union Bancaire Privee, said many investors underestimate Chinese AI models due to chip restrictions and skepticism. She added, "Alibaba’s Qwen 3.8 is another proof point, following Kimi K3," as evidence the performance gap is narrowing quickly [3].
Following the announcement, Alibaba’s shares surged by up to 7.3% in Hong Kong on August 3 [3, 4].
The AI landscape in China remains active: Moonshot AI released its Kimi K3 model with 2.8 trillion parameters in July 2026, and DeepSeek expanded access to its V4 Flash AI model on July 31 [1, 3, 4, 5, 6].
Alibaba plans to publicly release the Qwen3.8-Max model weights for download during the week of August 10, 2026 [1, 3, 4, 6].
