Project AURIONAI Use-Cases Research & Innovation
Google
visionfamily

Vision Transformer (ViT)

Google transformer image classifier that treats an image as a sequence of 16x16 patches; the base of most later vision transformers.

GoogleGoogleUSA
Figure from the Vision Transformer paper: the model splits an image into patches, embeds them and feeds them to a transformer encoder.
Levelfamily
Classvision
Parametersnot in the notes
Contextnot in the notes
Released22 Oct 2020
Opennessunknown
Runs onmobile or NPU, CPU server
Statusactive

Lineage

202020212022ViT-B/16ViT-B/16, Oct 2020ViT-L/16, Oct 2020ViT AugRegViT AugReg, Jun 2021
GoogleViT AugReg belongs to this familyvision18 Jun 202186.57M
GoogleViT-B/16 belongs to this familyvision22 Oct 202086M
GoogleViT-L/16 belongs to this familyvision22 Oct 2020304M
Where this comes from
Parent and child relations from model cards, Hugging Face base_model fields and the genealogy tables of the research notes. Confidence and evidence are the notes' own.

R&D Topics

Image classification backbone architectures (CNN and vision transformer)exemplar

Identity

Developer
Google
Family
not in the notes
Released
22 Oct 2020 sourcesannouncement page
Weights published
–
Record checked
12 Oct 2026
Where this comes from
Developer and dates from the research notes. Release date is the vendor's public announcement; weights date is when open weights were published.

Architecture and Size

Architecture
dense transformer
Total parameters
– sources
Active parameters
–
Checkpoint parameters
–
Note
ViT-Ti, S, B, L and H with 8, 14, 16 and 32 pixel patches; AugReg weights (arXiv 2106.10270).
Context length
not in the notes
Input
image
Output
labels and boxes
Where this comes from
Parameter counts as the developer states them. Where the checkpoint holds a different total, both are shown (register A16).

Openness and Access

Openness
unknown
Licence
not in the notes sources
Access routes
open weights download, cloud marketplace
Where this comes from
The licence file governs over registry tags and aggregator labels.

Deployment

Runs on
mobile or NPU, CPU server
Footprint
Edge or toolkit builds listed in: Qualcomm AI Hub, OpenVINO.
Runtimes
not in the notes
Where this comes from
Deployment tiers and runtimes as named in the notes. Footprint is the notes' figure, not a measurement by AURION.

Indonesia

Indonesian support
not stated
Regional languages
not in the notes
Jakarta hosting
unknown
Languages
not in the notes
Where this comes from
Indonesia fields are carried on every model; unknown is an allowed value.

Benchmark Results

The notes record no benchmark result for this model.

Industry Use Cases

No industry use case links this model.

Sources

  1. arxiv.org/abs/2010.11929 https://arxiv.org/abs/2010.11929
  2. github.com/GoogleCloudPlatform/vertex-ai-samples/tree/ma... https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden
  3. huggingface.co/google/vit-base-patch16-224 https://huggingface.co/google/vit-base-patch16-224
  4. aihub.qualcomm.com/models/vit https://aihub.qualcomm.com/models/vit
  5. github.com/pytorch/vision/blob/main/torchvision/models/v... https://github.com/pytorch/vision/blob/main/torchvision/models/vision_transformer.py
  6. huggingface.co/OpenVINO/vit_tiny_cls-fp16-ov https://huggingface.co/OpenVINO/vit_tiny_cls-fp16-ov
  7. huggingface.co/api/models?pipeline_tag=image-classification&... https://huggingface.co/api/models?pipeline_tag=image-classification&sort=downloads&limit=50

To cite this page: Project AURION, AI Use-Cases Research & Innovation, "Vision Transformer (ViT)", https://aurion.semai.id/models/google-vit/, records of October 2026.