Alibaba launches the latest generation flagship model preview version of Qwen3.5-Max-Preview

Alibaba's Tongyi Qianwen has released the Qwen3.5-Max-Preview version, showcasing significant improvements in code, mathematical, and multilingual understanding capabilities. As the flagship model in the Qwen3.5 series, it employs a hybrid expert (MoE) architecture with a total of 235B parameters and 22B activation parameters, achieving a performance breakthrough while maintaining efficient inference. The model supports context windows of up to 256K tokens and excels at complex inference and long text processing.