Type a sentence, move the camera, and the world keeps going. 🌍

Tencent's Hunyuan team has released Hy-Nomas, an interactive world model that builds a world as a stream of video segments. At each turn you can add a character, start an action or change the style with a free-form prompt, or steer the view with a camera path. Earlier segments stay in memory so the world stays consistent.

Output is 720p at 24 FPS playback. The standard setup uses eight GPUs, and a low-memory setup runs on a single RTX 4090. The release also includes cinematic and robotics LoRAs, training code and Hy-Nomas-Bench, a 280-case test set for text control, camera navigation and spatial memory.

Model files are open under Apache 2.0 on Hugging Face.

Full story: http://globalfeed.ai (link in bio)

#AI #WorldModel #OpenModels #VideoGeneration #Tencent #AINews #globalfeed