Alibaba's Qwen3.8-Max demonstrates competitive performance with aggressive pricing

The Qwen3.8-Max model with 2.4 trillion parameters achieved performance comparable to or exceeding rivals like Fable 5 and Kimi K3 on multiple benchmarks. On Code Arena, an independent evaluation for front-end web development, the model ranked fourth, behind Claude Opus 5, Kimi K3 Max, and Claude Opus 5 High. Alibaba's pricing ($2 per million tokens for input, $6 for output) positions itself among the most competitive on the market, intensifying the price war between AI labs. The company plans to release the model's open-source weights next week, differentiating itself from major American labs that maintain proprietary models. Experts like Andrew Yoon at CivAI caution against interpreting Chinese benchmarks hastily, arguing that recent models like Qwen 3.8 Max, despite being highly capable, still lag behind the American frontier.