Tencent Hunyuan and WorkBuddy joint team announced on September 7th that the Hunyuan Hy4preview model has been fully upgraded and launched. Since its initial release on WorkBuddy, the model has received a large amount of feedback from users and developers. The joint team has completed the first round of specialized iterations addressing the concentrated issues raised.
Regarding known issues in the previous version, such as long thinking and excessive self-verification in complex tasks, this upgrade has implemented specialized optimizations. To ensure that the optimization does not sacrifice task performance, the team used a dual monitoring mechanism of Bench metrics and human evaluation to verify the optimized version. The results showed that the model significantly reduced the number of task rounds and input/output token consumption while maintaining task performance.
This means that in actual Agent scenarios, the model's response efficiency and calling cost will be improved simultaneously.
The joint team stated that they will continue to iterate quickly and welcome users to continue experiencing and providing suggestions.
