StorySoftware
Hygon's Deep Computing 3 DCU completes adaptation for Tencent's 295B-parameter Hunyuan Hy3 preview
Hygon said its third-generation DCU now runs Tencent's Hunyuan Hy3 preview, a 295-billion-parameter model with a 256K-token context, after deep adaptation work. It shows a large frontier-class Chinese model running on domestic accelerators rather than Nvidia hardware.
- Model: Tencent Hunyuan Hy3 preview, 295B parameters.
- Supports 256K-token context.
- Targets reasoning, agent and code workloads.