Hy-Nomas
Tencent Hunyuan's open interactive video world model. It builds a world segment by segment from text and camera control.
Hy-Nomas generates a world segment by segment from multi-turn text instructions and camera control, at 720 by 1280 and 24 frames per second, with 8-step distilled inference. Weights are on Hugging Face under Apache 2.0. A single-GPU config targets an RTX 4090 with INT8 weights and CPU offload. Tencent also released Hy-Nomas-Bench, with 280 test cases. The technical report is coming soon.
Where it ranks
12 in Game dev
An open world model that extends a scene from text and camera moves.
See the game dev list15 in Image and video
Interactive video, segment by segment, from text and camera control.
See the image and video list
Pricing
Free
Apache 2.0 weights. Check the host you use for any extra cost.
Checked
Added October 26, 2026
Last verified October 28, 2026
Dated releases
Tencent released Hy-Nomas, an open interactive video world model, with Apache 2.0 weights.