Tencent's Open-Source World Model Hy-Nomas: 720p/24fps, Runs on a Single RTX 4090
mishig25 · x · 2026-10-24
Tencent Hunyuan released Hy-Nomas, a new open-source world model under Apache 2.0. Highlights: multi-turn text and camera control, continuous generation with KV cache memory, 720p/24 FPS output, and a single RTX 4090 inference option. Hugging Face staff ran early tests and found it very good. Permissive licensing plus consumer-GPU inference makes it notable for interactive world-model work.
More from Multimodal
- Moss image model natively generates consistent same-character comics in any style — AltruisticList6000 · 2026-10-24
- Fairground AI TV airs Halloween special with four AI-made horror films — DavidmComfort · 2026-10-24
- Dreamina's $0.29 plan actually includes 1,575 credits, a closer look at the deal — azed_ai · 2026-10-24
- AI-composed orchestra played by AI-animated paper puppets: full Claude + MiniMax pipeline — UnfortunateHurricane · 2026-10-24
- Turning Ideogram vector art into animation with Claude, a two-tool demo — azed_ai · 2026-10-24
- New talking portrait model P-Video-2-Pro-Avator wows with lip sync across ages and styles — azed_ai · 2026-10-24