Qianwen Office has launched a new “standard model” in terms of model supply. Based on this model, users complete the same tasks and achieve the same or even better task quality. The token generation speed is increased by 1 times, token consumption is reduced by half, and the experience is greatly improved. The improvement in user experience comes from model upgrades and collaborative optimization of models and agents. The standard model of Qianwen Office incorporates the Qwen 3.8-Flash model that has just been released. The model uses a new architecture, and the performance surpassed Claude Opus 4.6 with only 100 billion total parameters. At the same time, the Qianwen Big Model Team and the Qianwen Office team jointly designed it to provide an exclusive version of the Qwen3.8-Flash model for Qianwen Office, which was optimized during the model training and inference stages. According to tests with a large number of real office scenarios, using the standard model of the Thousand Questions Office, the token generation speed for a single task was increased by 1 times, and token consumption was reduced by an average of half.

Qianwen Office has launched a new “standard model” in terms of model supply. Based on this model, users complete the same tasks and achieve the same or even better task quality. The token generation speed is increased by 1 times, token consumption is reduced by half, and the experience is greatly improved. The improvement in user experience comes from model upgrades and collaborative optimization of models and agents. The standard model of Qianwen Office incorporates the Qwen 3.8-Flash model that has just been released. The model uses a new architecture, and the performance surpassed Claude Opus 4.6 with only 100 billion total parameters. At the same time, the Qianwen Big Model Team and the Qianwen Office team jointly designed it to provide an exclusive version of the Qwen3.8-Flash model for Qianwen Office, which was optimized during the model training and inference stages. According to tests with a large number of real office scenarios, using the standard model of the Thousand Questions Office, the token generation speed for a single task was increased by 1 times, and token consumption was reduced by an average of half.