AxForge
Sign in Get started

Model reference · open weights

Cosmos-Predict2-Video2World

Video nvidia Image→video 1 build Its own licence terms 192k dl/mo

Cosmos-Predict2-Video2World is an open-weight video model from NVIDIA. Cosmos-Predict2-2B-Video2World (BF16) weighs 14.2 GB; the smallest configuration that runs it is RTX 3060 12 GB.

  • Cosmos-Predict2-2B-Video2World is a 2B parameter diffusion transformer model by NVIDIA designed for image-to-video generation.
  • It takes a text description and an image as the first frame to predict future frames, producing 5-second video clips in mp4 format.
  • The model supports resolutions of 480P and 720P at 10 or 16 FPS, requires 32.54 GB of GPU VRAM, and is released under the NVIDIA Open Model License.

Summary of the nvidia/Cosmos-Predict2-2B-Video2World model card, 2026-10-01

Hugging Face ↗GitHub ↗

What it is

Released byNVIDIA
Released2025-04-25
VRAM14.2 GB for the weights

What it runs on

Memory and cards for Cosmos-Predict2-2B-Video2World (BF16)

14.2 GBweights, file size
9.7 GBbiggest part
537 MBruntime overhead
CardWeightsMemory
RTX 3060 12 GBtight (encoders offloaded)11.6 GB
RTX 4060 Ti 16 GBtight15.4 GB
RTX 3090 24 GBfits23.4 GB
RTX 4090 24 GBfits23.4 GB
RTX 5090 32 GBfits31.0 GB
L40S 48 GBfits44.0 GB
A100 80 GBfits78.2 GB
H100 80 GBfits78.1 GB
RTX PRO 6000 Blackwell 96 GBfits93.8 GB
DGX Spark (GB10) 128 GB unifiedfits107 GB
H200 141 GBfits138 GB
B200 180 GBfits176 GB

How it works

How video models work

Prompt / imagestart pointTemporal diffusionframes over timeVideoMP4 clipA video model generates a sequence of coherent frames from your prompt or a starting image.

Running it yourself

Run it on a rented GPU

Rent a machine by the hour. How to run this model is on its model card.

© 2026 AxForge · EU-hosted AI infrastructure Pricing Docs Trust Privacy Terms