Qwen 3.7 Max drops as frontier agent foundation

Qwen 3.7 Max targets the agentic era with a 1M context window and Sparse MoE architecture, providing the stability required for long-horizon autonomous tasks. Its integration of a deliberate Thinking Mode and a hybrid attention mechanism maintains high-fidelity reasoning and speed without performance degradation in complex tasks. These capabilities, combined with aggressive pricing, position it as a formidable high-performance alternative for multi-file engineering and agentic workflows.