Tencent introduces Hy4-flash AI model to reduce operational costs
Tencent has launched Hy4-flash, an AI model designed to lower the cost of routine office tasks such as data processing and document creation.

Tencent has introduced Hy4-flash, a new AI model optimized for standard workplace functions, including research, slide production, and spreadsheet management. Internal benchmarks conducted by the company indicate that the model reduces task costs by 95 percent compared to the full-scale Hy4 version.
With 224 billion parameters, the model supports multimodal inputs and a context window of one million tokens. In blind user testing, Hy4-flash demonstrated performance and satisfaction levels comparable to Claude Haiku 5.5, while offering improved completion times in the company’s internal metrics.
Hy4-flash is currently available via WorkBuddy and Yuanbao, with API access offered through Tencent Cloud’s TokenHub. The company has announced that the model will be free for WorkBuddy users until November 30, with future plans to release the model weights publicly.