After the reconstruction of Hunyuan, the first release of Hy3 is open source: the main focus is on comprehensive practicality, and the capabilities of the Agent have been greatly enhanced.

Tencent Hy3 preview language model is released and open-sourced. This is a hybrid expert model that combines quick and slow thinking, with a total of 295 billion parameters, 21 billion activation parameters, and a maximum support of 256K context length. This is the first model trained after the reconstruction of Hy3, and it is currently the most intelligent model in the Hy3 series, achieving significant improvements in complex reasoning, instruction following, context learning, code, intelligent agents, and reasoning performance.