Tencent open-sources Hy4 Preview: 770B params, 1M context, Apache 2.0
xeophon · x · 2026-08-28
Tencent has released Hy4 Preview on Hugging Face: 770B total parameters with 49B active (A49B), 1M context window, under Apache 2.0. Reported scores are strong, roughly on par with GLM-5.3. Commenters note Tencent has really picked up model training since Shunyu Yao joined. The chat template shows support for thinking mode, reasoning-effort control and tool calls.
Related event: Tencent Open-Sources Hunyuan Hy4 Preview: 770B MoE with 1M Context(28 posts)→
More from Models
- Release season returns: Nemotron 3.5, FLUX 3 Klein, StepFun 5 and more in one day — TheZachMueller · 2026-10-16
- Rumor: OpenAI has a model better than GPT 6, possible late-October release — ChrisGPT · 2026-10-16
- Meta open-weights 816B Muse Spark Open, Apple tunes it for MLX on Mac — rohanpaul_ai · 2026-10-16
- GPT-6 Dynamic UI called an extraordinary math tutor with 'no comparison' — AarushSelvan · 2026-10-16
- Clean model weights could be 'a nuclear bomb' for continual learning — mayfer · 2026-10-16
- Running Nemotron-3.5-Bolt-74B MoE on a single RTX 3060 with 12GB VRAM — cephaloform · 2026-10-16