# Tencent releases Hy-Nomas, a free-form interactive video world generation model

> Original title: tencent/Hy-Nomas

- Source: Tencent · new models on Hugging Face
- Published: 2026-10-21T02:17:37.000Z
- Bad Influence: https://badinfluence.news/items/v82c6eetzfx7yz6vuuzu9epap
- Original: https://huggingface.co/tencent/Hy-Nomas

## Summary

Tencent releases Hy-Nomas on Hugging Face, a video world generation model that takes text instructions and camera control across multi-turn interaction. It generates one video segment at a time at 720 × 1280 and 24 FPS using 8 denoising steps, and reuses context through a KV cache. A single-GPU configuration on an NVIDIA GeForce RTX 4090 is described, and Tencent also releases the Hy-Nomas-Bench suite of 280 test cases. The model is licensed under Apache 2.0.

## Why it matters

The model card specifies its interaction design, segment-based architecture, and single-GPU configuration, which helps readers judge what it can run on.