Project AURIONAI Use-Cases Research & Innovation
Google
multimodal LLMfamily

PaliGemma 2

PaliGemma upgraded to Gemma 2 language models in 3B, 10B and 28B sizes at three image resolutions, for transfer to OCR, captioning and medical imaging tasks.

GoogleGoogleUSA
Figure from the PaliGemma 2 paper: example of referring segmentation output for a prompt about two puffins.
Levelfamily
Classmultimodal LLM
Parametersnot in the notes
Contextnot in the notes
Released21 Nov 2024
Opennessopen weights restricted
Runs onnot in the notes
Statusactive

Lineage

20242025PaliGemma 2 10B pt 448PaliGemma 2 10B pt 448, Nov 2024PaliGemma 2 3B pt 224, Nov 2024PaliGemma 2 28B pt 896, Nov 2024
GooglePaliGemma 2 28B pt 896 belongs to this familymultimodal LLM22 Nov 202427.65B
GooglePaliGemma 2 10B pt 448 belongs to this familymultimodal LLM21 Nov 20249.66B
GooglePaliGemma 2 3B pt 224 belongs to this familymultimodal LLM21 Nov 20243.03B
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

Document, OCR and chart understandingproduced byVisual instruction tuningproduced by

Identity

Developer
Google
Family
not in the notes
Released
21 Nov 2024 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
–
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 restricted
Licence
Gemma Terms of Use 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/2412.03555 https://arxiv.org/abs/2412.03555
  2. github.com/GoogleCloudPlatform/vertex-ai-samples/tree/ma... https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden
  3. docs.roboflow.com/models/supported-models/paligemma2.md https://docs.roboflow.com/models/supported-models/paligemma2.md
  4. huggingface.co/google/paligemma2-3b-pt-224 https://huggingface.co/google/paligemma2-3b-pt-224
  5. huggingface.co/google/paligemma2-10b-pt-448 https://huggingface.co/google/paligemma2-10b-pt-448

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