Key facts
- Alibaba launched its Qwen3.8 Max AI model, described as its most powerful.
- The model is a 2.4-trillion-parameter sparse Mixture-of-Experts preview.
- Alibaba claimed the model ranked second only to Anthropic's Claude Fable 5, but provided no independent benchmarks.
- The model supports multimodal inputs and has a 983,616-token context window.
- Alibaba was designated a Chinese Military Company by the Pentagon in June 2026.
- Benchmark results indicate the Qwen model underperforms several domestic and international competitors.
Alibaba officially launched its Qwen3.8 Max AI model, which it described as its most powerful to date. However, benchmark performance results suggest the model falls short of the company's claims, including a statement that it ranked second only to Anthropic's Claude Fable 5. This claim was made without accompanying benchmark data or model cards, raising questions about its validity.
The Qwen3.8 Max-Preview is a 2.4-trillion-parameter sparse Mixture-of-Experts model, accessible through Alibaba's Token Plan subscription at a discounted rate. It supports multimodal inputs—text, images, video, and documents—and features a large context window of 983,616 tokens with a maximum output length of 131,072 tokens. The model's thinking capability is active by default with three intensity settings.
Key technical details such as the active-parameter count, open-weight release date, and standard pay-as-you-go API pricing remain unconfirmed. The model also lacks a listing on independent leaderboards like Hugging Face. Adding to the scrutiny, Alibaba was designated a Chinese Military Company by the Pentagon in June 2026, a classification it is contesting in court. The company's adherence to China's National Intelligence Law further complicates adoption decisions for enterprises considering the model.