Project AURIONAI Use-Cases Research & Innovation
Google
embeddingfamily

SigLIP 2

Multilingual image-text encoders trained with a sigmoid loss plus captioning and self-supervised objectives, for zero-shot classification, retrieval and as VLM vision encoders.

GoogleGoogleUSA
Table of SigLIP 2 evaluation results across model sizes and resolutions, from the model card.
Levelfamily
Classembedding
Parametersnot in the notes
Contextnot in the notes
Released17 Feb 2025
Opennessopen weights permissive
Runs onnot in the notes
Statusactive

Lineage

20252026SigLIP 2 base patch16 224SigLIP 2 base patch16 224, Feb 2025SigLIP 2 giant patch16 384, Feb 2025SigLIP 2 So400m patch14 384, Feb 2025
GoogleSigLIP 2 base patch16 224 belongs to this familyembedding17 Feb 2025375.19M
GoogleSigLIP 2 giant patch16 384 belongs to this familyembedding17 Feb 20251.87B
GoogleSigLIP 2 So400m patch14 384 belongs to this familyembedding17 Feb 20251.14B
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

Contrastive image-text encoders (CLIP and SigLIP lines)exemplar

Identity

Developer
Google
Family
not in the notes
Released
17 Feb 2025 sourcesregistry date
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-B (86M) to g (1B); NaFlex versions keep native aspect ratio. The first SigLIP 400M encoder is a separate AURION record (siglip-400m).
Context length
not in the notes
Input
text, image
Output
text
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
open weights permissive
Licence
Apache License 2.0 sources
Access routes
open weights download
Where this comes from
The licence file governs over registry tags and aggregator labels.

Deployment

Runs on
not in the notes
Footprint
not in the notes
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/2502.14786 https://arxiv.org/abs/2502.14786
  2. github.com/google-research/big_vision https://github.com/google-research/big_vision
  3. aihub.qualcomm.com/models/siglip2 https://aihub.qualcomm.com/models/siglip2
  4. huggingface.co/api/models?pipeline_tag=zero-shot-image-class... https://huggingface.co/api/models?pipeline_tag=zero-shot-image-classification&sort=downloads&limit=50
  5. huggingface.co/google/siglip2-base-patch16-224 https://huggingface.co/google/siglip2-base-patch16-224
  6. huggingface.co/google/siglip2-so400m-patch14-384 https://huggingface.co/google/siglip2-so400m-patch14-384

To cite this page: Project AURION, AI Use-Cases Research & Innovation, "SigLIP 2", https://aurion.semai.id/models/google-siglip-2/, records of October 2026.