Nov 1, 2026

Tencent's Hy-Nomas: an open-source world you can steer

Tencent Hunyuan released Hy-Nomas, an open-source interactive video world model you steer with text prompts or your keyboard. There are no benchmark scores or tech report yet.

Sir Toastimir

Footage: Tencent Hunyuan

What it is

Hy-Nomas is an open-source interactive video world model from Tencent Hunyuan. Instead of making one fixed clip, it generates video one segment at a time, and you decide what happens next.

How you steer it

You can type a text instruction for the next part of the scene, or control the camera with keyboard inputs, and switch between the two. Tencent's model card lists output at 720x1280 resolution and 24 frames per second.

Robots

There is an optional robotics LoRA (a small add-on) trained on a real-world robotic-arm dataset, for scenes of robot arms doing tasks.

What you need to run it

The standard configuration uses 8 GPUs. Tencent also lists a reduced configuration that runs on a single RTX 4090. The code and weights are released under the Apache 2.0 license, and there is an online interactive demo.

What we don't know yet

Tencent hasn't published benchmark scores or a technical report yet (the README says the report is coming soon), so this post only describes what the model is built to do.

Visuals in the video

All footage is from Tencent Hunyuan's official Hy-Nomas promo video.