Today, Alibaba’s Qwen Qwen3.8-Max artificial intelligence model was officially launched, with 2.4 trillion total parameters.
According to the official announcement, Qwen3.8-Max delivers comprehensive improvements in programming, real-world office work, long-horizon tasks, and multimodal agents. It can complete complex, open-ended objectives end to end with minimal human intervention and deliver reliable results.
In the authoritative third-party Arena leaderboard released today, Alibaba’s Qwen models ranked just behind Anthropic’s Claude series, placing their overall performance among the world’s leading large models. The Qwen3.8 API is now available on the Qwen AI platform and has also been integrated into “Qwen Office,” an agent product launched by Alibaba today.
The Qwen 3.8-MAX model weights will be open-sourced next week, along with Qwen3.8-27B. Developers can access Qwen3.8 API services through the Qwen AI platform. In China, the price is 12 yuan per million input tokens and 36 yuan per million output tokens, while implicit cache hits cost only 1.5 yuan. International input and output prices are 40% and 24% of Opus 5’s prices, respectively.