Qianwen Office launches Qwen3.8-Flash, with generation speed increased by 100% and token consumption reduced by 75%.

On August 26 evening, Qianwen Office launched the newly released Qwen3.8-Flash model for the first time, while also introducing a standard mode.

Starting today, all users can experience Qwen3.8-Flash through the new standard mode. Based on the latest model, users can complete tasks with fewer credit consumption and faster token throughput.

Going forward, Qianwen Office's model offerings will only include standard and premium modes. 95% of daily tasks can be completed through the standard mode, with only 5% of complex tasks requiring the premium mode.

The improvement in user experience comes from model upgrades and Agent collaborative optimization. The newly architected Qwen3.8-Flash achieves performance surpassing Claude Opus 4.6 with a total parameter scale in the hundreds of billions.

Meanwhile, the Qianwen large model team and the Qianwen Office team have jointly launched an office-specific version of Qwen3.8-Flash, with specialized training and tuning for scenarios such as multi-step planning, tool selection, and context compression, further improving throughput efficiency through inference optimization and a customized Harness architecture.

In real office scenario testing, the single-task generation speed of Qianwen Office's standard mode has increased by approximately 100%, with token consumption reduced by an average of 75%.

In real AI application scenarios, high performance typically means high cost and high latency, while low cost requires sacrificing intelligence. The deep collaborative optimization between Agent and model is breaking the "impossible triangle" of performance, cost, and speed.

As model intelligence density continues to rise, and with the bidirectional optimization between Qianwen Office and the model, Agent is about to bid farewell to token anxiety and usher in an era of "plenty and affordable."