The Tencent Hy Team has open-sourced Hy4 preview, a new-generation Mixture-of-Experts flagship model comprising 770B total parameters with 49B activated per token. The architecture features 78 layers where the first uses a dense FFN and the remaining 77 use MoE with 256 routed experts, alongside a native MTP layer for speculative decoding.
- Hy4 preview is trained on data from Tencent software engineers, game developers, and finance analysts to improve productivity tasks.
- In blind side-by-side evaluations on 203 engineering tasks by 163 internal experts, Hy4 preview scored 2.99, slightly ahead of GLM 5.3 (2.92) and Kimi K3 (2.94).
- The model supports deployment via vLLM or SGLang with MTP enabled, and includes an AngelSlim toolkit for compression.
The release aims to push the boundaries of model intelligence while making it more abundant and affordable through early iteration based on user feedback.