Tencent Hunyuan Hy4 Preview: 770B-Parameter Open Model With 49B Active and Native 1M Context
TencentHunyuan · x · 2026-09-09
Tencent Hunyuan amplified a developer demo of Hy4 preview: an open model with 770B total parameters, 49B active per token, native 1M-token context, Apache 2.0 license, official FP8 weights, and direct vLLM and SGLang support.
In the demo, Hy4 generated a colorful, playable 2D space shooter from a single prompt, complete with controls, enemies, particle effects and score tracking. The author notes it's still a preview and recommends benchmarking tokens/s against BF16/FP8 builds.
More from Models
- Heavy user on Astra: beats a junior hire on cost, but still fumbles simple tasks — RachelVT42 · 2026-09-09
- OpenAI Claims Its Agents Solved the Navier-Stokes Millennium Prize Problem — JosephJacks_ · 2026-09-09
- As Models Master Structured Tasks, Creative Writing Keeps Getting Worse — teodorio · 2026-09-09
- Math benchmark success shows log-linear diminishing returns with test-time compute, says ramez — sebkrier · 2026-09-09
- An image prompt carries about as much information as taking a photo, argues Toby Ord — tobyordoxford · 2026-09-09
- Toby Ord: image models are like a lossy compression format for photos — tobyordoxford · 2026-09-09