【#Tech24H】QwenWork has introduced a new “Standard Mode” for model provisioning. With this mode, users can accomplish the same tasks with equal or even better quality, while doubling token generation speed and cutting token consumption by half, significantly improving the user experience. The Standard Mode is powered by the newly released Qwen3.8Flash model, which uses a new architecture and, with only around 100 billion total parameters, already outperforms Claude Opus 4.6. In addition, the Qwen model team and the QwenWork team have jointly designed an exclusive version of Qwen3.8Flash for Qwen Office, with optimisations at both the training and inference stages. Tests in a wide range of realworld office scenarios show that using QwenWork Standard Mode doubles token generation speed per task and reduces average token consumption by half. In the future, 95% of daily tasks will be handled by the Standard Mode, with only 5% of complex tasks requiring the Advanced Mode. This significantly lowers the cost barrier for highfrequency Agent usage scenarios.[ By Zhang Liyan | Tang Ruohan ]