Vision Transformers — EchtRaaW Passport

// PASSPORT

Vision Transformers

Google DeepMind · Multimodal model · Registered 2026-03-06

IDENTITY

Developer

Google DeepMind

Confirmed by

Confirmed via https://huggingface.co/blog/vision-transformers

First registered

2026-03-06 (rss_pipeline)

Developer confirmed

2020-10-22

Gate

PROVENANCE COMPLETE

PROVENANCE

Claimed developer

Google DeepMind

Confirmed developer

Google DeepMind

Delta

— · match confirmed

EU COMPLIANCE

AI Act

unknown

PASSPORT FIELDS

Model type

Multimodal

Parameters

86M (ViT-Base); variants range from ~86M to ~632M (ViT-Large, ViT-Huge)

Context window

Unknown

Open source

Yes

Weights available

Yes

License

Apache-2.0

PROVENANCE TIMELINE

2026-03-06

First recorded in AI news