Model reference · open weights

dinov3-vits16plus-pretrain-lvd1689m

dinov3-vits16plus-pretrain-lvd1689m is an open-weight embedding model from Meta. dinov3-vits16plus-pretrain-lvd1689m (FP32) weighs 57 MB; the smallest configuration that runs it is RTX 3060 12 GB.

• DINOv3-vits16plus-pretrain-lvd1689m is a 29M parameter Vision Transformer model developed by Meta for image feature extraction.
• It uses a patch size of 16 and was pretrained on the LVD-1689M web dataset to produce dense features for downstream tasks such as classification and segmentation.
• The model is released under a custom license and supports English.

Summary of the facebook/dinov3-vits16plus-pretrain-lvd1689m model card, 2026-10-01

Hugging Face ↗Paper ↗