FlashNEWS BRIEFING

Tencent Hunyuan Open-Sources Interactive Video World Generation Model Hy-Nomas

1 min read

88Z AI, 26 October — Tencent Hunyuan has released Hy-Nomas, an open-source interactive video world generation model. It supports text and camera trajectory control while maintaining long-range consistency. Optimized for local inference on a single RTX 4090 with LoRA fine-tuning support, it is suitable for gaming, simulation, and film pre-visualization.

At a glance

  • Supports text and camera trajectory control, using lifelong KV cache for long-range consistency
  • Optimized for local inference on a single RTX 4090, with LoRA fine-tuning support
Cover image for Tencent Hunyuan Open-Sources Interactive Video World Generation Model Hy-Nomas
Cover image for Tencent Hunyuan Open-Sources Interactive Video World Generation Model Hy-Nomas

The update

Tencent Hunyuan has officially open-sourced Hy-Nomas, an interactive video world generation model. This model offers robust control capabilities, allowing video generation to be guided by text descriptions and camera trajectories, while supporting continuous generation. To address frame flickering in long videos, Hy-Nomas introduces lifelong KV cache technology, effectively maintaining consistency over long time ranges. In terms of deployment efficiency, the model has been optimized through distillation and INT8 quantization, significantly lowering hardware requirements. Users can perform local inference with just a single RTX 4090 GPU. Furthermore, Hy-Nomas supports LoRA fine-tuning, enabling developers to customize it for specific needs. Its applications are extensive, covering game prototyping, robot simulation training, and film pre-visualization, demonstrating significant commercial potential.

Sources & editorial note

This article is an AI-assisted synthesis by 88Z AI based on the public sources below. Product claims and availability are attributed to their publishers. It does not represent independent testing or first-hand reporting.

  1. Hy-NomasSource published

Source dates above refer to the original announcements; the date under the headline is this article’s publication on 88Z AI.

Related reading

All stories
88Z.ai