embedding

EmbeddingGemma 2

Google's embedding model on the Gemma 4 architecture, 768-dimension Matryoshka (512, 256, 128).

GoogleUSA

Levelmodel

Classembedding

Parameters740M

Context8,192

Released7 Oct 2026

Opennessopen weights permissive

Runs onnot in the notes

Statusactive

Benchmark Results

vendor-reported

MTEB Codeembeddings

vendor-reported

78.68

04080

Score in score. Higher is better.

huggingface.co

MTEB English v2embeddings

vendor-reported

68.46

04080

Score in score. Higher is better.

huggingface.co

MTEB multilingual v2embeddings

vendor-reported

61.36

04080

Score in score. Higher is better.

huggingface.co

Where this comes from

Every score names who ran it: vendor-reported, maintainer-run or independent. A rank is shown with the population it was ranked in (gap rule 3).

R&D Topics

Retrieval-augmented generation (RAG)used by Text embeddings for semantic searchproduced by

Identity

Developer: Google
Family: not in the notes
Released: 7 Oct 2026 sourcesregistry date
Weights published: –
Record checked: 10 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: not in the notes Gemma 4 architecture
Total parameters: 740M sources
Active parameters: –
Checkpoint parameters: –
Note: 740M total (270M text path; vision and audio encoders)
Context length: 8,192 tokens
Input: text, image, audio
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: 100+ languages

Where this comes from

Indonesia fields are carried on every model; unknown is an allowed value.

Lineage

The notes record no lineage for this model.

Industry Use Cases

No industry use case links this model.

Links

Developer

Googlebig tech, USA

Model

EmbeddingGemma 2Google, embedding

R&D TopicsRetrieval-Augmented Generation (RAG)Text Embeddings for Semantic Search

Use CasesEmbeddings and Retrieval for SEA LanguagesPublic SectorUse cases of models in the same R&D topic; none is recorded for this model itself.

Who Runs Itnone recordedDeployments of these use cases; none names this model.

Graph

Organisations 1

Googlebig tech, USADeveloper

R&D topics 2

Retrieval-Augmented Generation (RAG)Grounding generation in retrieved documents.used by Text Embeddings for Semantic SearchTraining encoders that turn text in many languages into vectors so that search, clusterin…produced by

Sources

google/embeddinggemma-2 , Hugging Face huggingface.co https://huggingface.co/google/embeddinggemma-2

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