Difix3D+
No estimate
No hardware requirements for this model
This model's weights are open, but no parameter count has been published for it. Every memory and speed figure starts from that number, so we would rather show nothing than a fabricated estimate.
On record
Full specification
Everything on record for this model. Most of it describes how it was trained rather than how it runs — useful context for judging how much work went into it, and how it compares with models built at a different scale.
Origin
Who built this model, where, and when it was published.
- Organisation
- NVIDIA,National University of Singapore,University of Toronto,Vector Institute
- Organisation type
- Industry,Academia,Academia,Academia
- Country
- United States of America, Singapore, Canada
- Published
- 3 March 2025
- Authors
- Jay Zhangjie Wu, Yuxuan Zhang, Haithem Turki, Xuanchi Ren, Jun Gao, Mike Zheng Shou, Sanja Fidler, Zan Gojcic, Huan Ling
What it does
The problem areas the model was built for. A model can carry several of each.
- Domain
- 3D modeling
- Task
- 3D reconstruction
- Base model
- SD-Turbo
Size
How large the model is and how much data it was trained on. Parameters are the figure that decides whether it fits on a given graphics card.
- Training data
- tokens
Availability
Whether you can obtain the model and run it on your own hardware, which is what decides if any of the graphics-card figures on this page apply.
- Weights
- Open — downloadable
- Model access
- Open weights (non-commercial)
- Training code
- Open (non-commercial)
- Hugging Face
- nvidia
NVIDIA License https://huggingface.co/nvidia/difix "This model is ready for research and development/non-commercial use only." NVIDIA non-commercial license for training code: https://github.com/nv-tlabs/Difix3D/tree/main
How it is classified
Labels the source dataset applies when tracking notable models, and how confident it is in the entry.
- Record confidence
- Unknown
Sources
Where this record came from and when it was last checked.
- Reference
- Difix3D+: Improving 3D Reconstructions with Single-Step Diffusion Models
- Last updated
- 28 November 2025
What the numbers mean
About this model
Difix3D+ was published by NVIDIA,National University of Singapore,University of Toronto,Vector Institute, in United States of America, in March 2025. industry,Academia,Academia,Academia is the category the publisher falls under.
It works in 3D modeling, and is recorded as doing 3D reconstruction.
It is derived from SD-Turbo rather than trained from scratch, which is the usual way a specialised model is produced.
Published weights mean the model runs on your machine rather than someone else's, which is what makes the hardware question below answerable at all. It is published under the nvidia organisation on Hugging Face.
Answers
Difix3D+ — common questions
Where can I download Difix3D+?
Its weights are published under the nvidia organisation on Hugging Face. We do not host model files — this site calculates what hardware is needed to run them.
What GPU do I need to run Difix3D+?
We cannot say. Difix3D+ has open weights, but no parameter count has been published for it, and every memory and speed calculation starts from that number. We would rather show nothing than a fabricated estimate.
Is Difix3D+ open source?
Its weights are published, so Difix3D+ can be downloaded and run on your own hardware. Note that open weights is not the same as open source in the full sense — it says nothing about the training data, the training code, or the commercial terms attached.
How many parameters does Difix3D+ have?
No parameter count has been published for Difix3D+, which is why no memory or speed figure appears on this page.
Who created Difix3D+?
Difix3D+ was published by NVIDIA,National University of Singapore,University of Toronto,Vector Institute, based in United States of America, categorised as industry,Academia,Academia,Academia.
When was Difix3D+ released?
Difix3D+ was published in March 2025.
What is Difix3D+ used for?
Difix3D+ works in 3D modeling, and is recorded as handling 3D reconstruction. A model can carry several of each, so these are the areas it was built for rather than a limit on what it will attempt.
The other direction
Looking at it from the other side?
This page starts from the model. If you already own a card and want to know everything it will run, start from the hardware instead.