EmbeddingGemma 2: Unified and Efficient Multimodal Representations

We present EmbeddingGemma 2, an open multimodal embedding model which natively represents text, images, audio, video, and combinations thereof in a unified space. Our approach introduces improvements such as simplified per-layer embeddings and a multi-stage curriculum with modality and objective progression which ensure strong quality across unimodal, cross-modal, and mixed-modal evaluations. Notably, in comprehensive embedding benchmarks across all modalities, EmbeddingGemma 2 establishes a new Pareto frontier, outperforming both specialist models and models two to twelve times larger. Furthermore, our design offers flexibility for resource-constrained environments, supporting selective modality encoder loading, adjustable vision token budgets, truncatable embedding dimensions, and quantized weights. We present ablations analyzing key decisions and release EmbeddingGemma 2 to promote further research.

Authors

Publication Details

Journal
arXiv (Cornell University)
Published
2026-10-27
DOI
https://doi.org/10.48550/arxiv.2610.37364
Primary Topic
Multimodal Machine Learning Applications
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
preprint

EmbeddingGemma 2: Unified and Efficient Multimodal Representations

Andrés Sevilla, Jae Hyeon Yoo, Sahil Dua, Gustavo Hernández Ábrego et al.
arXiv (Cornell University)
Multimodal Machine Learning Applications
preprint

EmbeddingGemma 2: Unified and Efficient Multimodal Representations

Andrés Sevilla, Jae Hyeon Yoo, Sahil Dua, Gustavo Hernández Ábrego, Georgios Kathareios, Madhuri Shanbhogue, Rui Meng, Mogan Shieh, Chandramouli Amarnath, Ibrahim Alabdulmohsin, Cormac Brick, Tanmaya Shekhar Dabral, Tom Duerig, Pooja Consul, Alberto Montes, Luciano Martins, Babak Samari, Mojtaba Seyedhosseini, Brendan McCloskey, Daniel Salz, Daniel M. Cer, Keegan Mosley, Joe Zou, Iván Llanos, Edouard Yvinec, Omar Sanseviero, I. A. Ballantyne, Philippov Ilya, Tris Warkentin, Josiah G. Aklilu, Jovan Andonov, Yi-Chun Kuo, Renjie Wu, Kat Black, Armand Joulin, Kendra Byrne, Glenn Cameron, Henrique Schechter Vera, Sindhu Raghuram Panyam, Sara Smoot, Brian Potetz, Yichang Chen, Shanfeng Zhang, Hesen Zhang, Hengxuan Ying, Yunan Zhang, Sebastian Russo, Olivier Lacombe, Juyeong Ji, Maarten Grootendorst, Wai Hon Law, Jiaxing Wu, Matt Kreileder, Zhongli Ding, Qin Yin, Weiyi Wang, Paul Suganthan, Ji Ma, Kristen Quan, Matthew Soulanille, Yu-Hui Chen, Alice Zheng, Andreas Hess, Shih-Cheng Huang, Zhe Li, Doug Reid, Max Matveev, Shen Gao, Xu Chen, Shengcao Cao, Suril Shah, Vinod Mamillapalli, Yi-Ting Chen
preprint en

Abstract

We present EmbeddingGemma 2, an open multimodal embedding model which natively represents text, images, audio, video, and combinations thereof in a unified space. Our approach introduces improvements such as simplified per-layer embeddings and a multi-stage curriculum with modality and objective progression which ensure strong quality across unimodal, cross-modal, and mixed-modal evaluations. Notably, in comprehensive embedding benchmarks across all modalities, EmbeddingGemma 2 establishes a new Pareto frontier, outperforming both specialist models and models two to twelve times larger. Furthermore, our design offers flexibility for resource-constrained environments, supporting selective modality encoder loading, adjustable vision token budgets, truncatable embedding dimensions, and quantized weights. We present ablations analyzing key decisions and release EmbeddingGemma 2 to promote further research.

arXiv (Cornell University)
Multimodal Machine Learning Applications
DOIOpen Access PDFOpenAlex
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon