Runs On
What it takes to serve dinov3-vitl16-pretrain-lvd1689m (303M parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them.
Memory is the weights at that precision plus 20% for the runtime and a short context; a long context needs more. Prices are the lowest on-demand hourly rates in the SAVRN Index, read Sep 18, 2026.
SAVRN's Notes on dinov3-vitl16-pretrain-lvd1689m
What does a vision model that returns features instead of answers earn its rack space with? For dinov3-vitl16-pretrain-lvd1689m the answer is embeddings: it turns a picture into a feature vector that a downstream classifier or search index consumes. At 303 million parameters and 24 layers, the 16-bit weights are 0.6 GB and need 0.7 GB of memory; 8-bit drops that to 0.4 GB. One MI300X at $1.85 an hour is the cheapest card in our table and far more than the model needs, so the decision is about images per hour, not fitting weights.
The license field reads other with no summary attached, and the listed publisher is Camenduru while the weights derive from facebook/dinov3-vit7b16-pretrain-lvd1689m, so trace the terms to that source before any commercial deployment. The paper is arXiv:2508.10104. There is no context length to size, so your check is image throughput per card.
SAVRN Research, 2026-09-18
Model Card
DINOv3 is a family of versatile vision foundation models that outperforms the specialized state of the art across a broad range of settings, without fine-tuning. DINOv3 produces high-quality dense features that achieve outstanding performance on various vision tasks, significantly surpassing previous self- and weakly-supervised foundation models. These are Vision Transformer and ConvNeXt models trained following the method described in the DINOv3 paper. 12 models are provided: - 10 models pretrained on web data (LVD-1689M dataset) - 1 ViT-7B trained from scratch, - 5 ViT-S/S+/B/L/H+ models distilled from the ViT-7B, - 4 ConvNeXt-{T/S/B/L} models distilled from the ViT-7B, - 2 models…
Excerpt from the card by Camenduru, licensed other.
Configuration
- Architecture
- DINOv3ViTModel
- Layers
- 24
- Hidden size
- 1,024
- Feed-forward size
- 4,096
- Attention heads
- 16
- RoPE base
- 100
- Stored precision
- float32
- Model type
- dinov3_vit
Identity and Version
- Repository
- camenduru/dinov3-vitl16-pretrain-lvd1689m
- Publisher
- Camenduru
- Task
- Image feature extraction
- Modality
- Other
- Library
- transformers
- Parameters
- 303M parameters
- Languages
- en
- Revision
- 3c276edd87d6f6e569ff0c4400e086807d0f3881
- First published
- 2025-12-17
- Last updated
- 2025-12-17
Files and Weights
6 files, 1.2 GB in total. The weights are 1 file totalling 1.2 GB in safetensors.
Weights1 file · 1.2 GB
Configuration2 files · 1.3 KB
Documentation2 files · 22.0 KB
Repository1 file · 1.5 KB
Every file
License and Download
- License
- other
- Access
- Open weights, no gate
- Download size
- 1.2 GB
Built From
- Derived from
facebook/dinov3-vit7b16-pretrain-lvd1689m
- Described by
arXiv:2508.10104
Memory Requirements
Weights only, from the published parameter count; the key-value cache and runtime add to this.
Compare dinov3-vitl16-pretrain-lvd1689m
Questions About dinov3-vitl16-pretrain-lvd1689m
How much GPU memory does dinov3-vitl16-pretrain-lvd1689m need?
About 0.7 GB at 16-bit and 0.2 GB at 4-bit: the weights (303M parameters) plus a working margin. A long context needs more.
What is the cheapest GPU to run dinov3-vitl16-pretrain-lvd1689m on?
At 16-bit, 1x MI300X from $1.85 an hour; at 4-bit, 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists.
What license is dinov3-vitl16-pretrain-lvd1689m released under?
other, as its publisher declares it. Read the license text before commercial use.
Similar Models
Vision Transformer (ViT) model trained using the DINOv2 method. It was introduced in the paper DINOv2: Learning Robust Visual Features without Supervision by Oquab et al. and first released in this repository. Disclaimer: The team releasing DINOv2 did not write a model card for this model so this model card has been written by the Hugging Face team. The Vision Transformer (ViT) is a transformer encoder model (BERT-like) pretrained on a large collection of images in a self-supervised fashion. Images are presented to the model as a sequence of fixed-size patches, which are linearly embedded. One also adds a [CLS] token to the beginning of a sequence to use it for classification tasks. One…
Open weights
apache-2.0
304M parameters
transformers
Vision Transformer (ViT) model trained using the DINOv2 method. It was introduced in the paper DINOv2: Learning Robust Visual Features without Supervision by Oquab et al. and first released in this repository. Disclaimer: The team releasing DINOv2 did not write a model card for this model so this model card has been written by the Hugging Face team. The Vision Transformer (ViT) is a transformer encoder model (BERT-like) pretrained on a large collection of images in a self-supervised fashion. Images are presented to the model as a sequence of fixed-size patches, which are linearly embedded. One also adds a [CLS] token to the beginning of a sequence to use it for classification tasks. One…
Open weights
apache-2.0
87M parameters
transformers
Model · Image feature extraction
Google
Vision Transformer (ViT) model pre-trained on ImageNet-21k (14 million images, 21,843 classes) at resolution 224x224. It was introduced in the paper An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale by Dosovitskiy et al. and first released in this repository. However, the weights were converted from the timm repository by Ross Wightman, who already converted the weights from JAX to PyTorch. Credits go to him. Disclaimer: The team releasing ViT did not write a model card for this model so this model card has been written by the Hugging Face team. The Vision Transformer (ViT) is a transformer encoder model (BERT-like) pretrained on a large collection of images in a…
Open weights
apache-2.0
86M parameters
transformers
Vision Transformer (ViT) model trained using the DINOv2 method. It was introduced in the paper DINOv2: Learning Robust Visual Features without Supervision by Oquab et al. and first released in this repository. Disclaimer: The team releasing DINOv2 did not write a model card for this model so this model card has been written by the Hugging Face team. The Vision Transformer (ViT) is a transformer encoder model (BERT-like) pretrained on a large collection of images in a self-supervised fashion. Images are presented to the model as a sequence of fixed-size patches, which are linearly embedded. One also adds a [CLS] token to the beginning of a sequence to use it for classification tasks. One…
Open weights
apache-2.0
22M parameters
transformers
A Vision Transformer (ViT) image feature model. Pretrained on LVD-142M with self-supervised DINOv2 method. - An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale: https://arxiv.org/abs/2010.11929v2 Explore the dataset and runtime metrics of this model in timm model results.
Open weights
apache-2.0
22M parameters
timm
