千问办公首发上线Qwen3.8-Flash，生成速度提升100%，Token消耗减少75%

TL;DR - Qwen Office launched Qwen3.8-Flash as its new standard model, claiming roughly 2× faster generation and 75% lower average token usage in real office tasks. The release combines a specialized model variant with agent and inference-stack optimizations to improve performance, cost, and latency.

• Qwen3.8-Flash was tuned for multi-step planning, tool selection, and context compression in office workflows.
• Further gains come from inference optimizations and a customized agent harness architecture.
• Qwen Office says its standard mode can handle 95% of everyday tasks, reserving advanced mode for the remaining 5% of complex work.
• The reported speed and token-efficiency improvements are based on the company’s internal real-world office scenario tests.