Paper

Gemini Embedding: Generalizable Embeddings from Gemini

Chen, Yichang, Duerig, Tom, Ye Xia, Sung, Yunhsuan, Dong, Zhe, Goenka, Sonam, Salz, Daniel, Baumgartner, Simon, Chen, Blair, Chen, Kaifeng, Suganthan, Paul, Frank Palma Gomez, Gill, Karan, Zhou, Wenlei, Hoffmann, Raphael, Seyedhosseini, Mojtaba, Boratko, Michael, Gleicher, Zach, Dua, Sahil, Doumanoglou, Andreas, Choi, Min, Parashar Shah, Huang, Shuo, Li, Zhe, Lee, Jinhyuk, Gustavo Hernández Ábrego, Ghiya, Rakesh, Sai Meher Karthik Duddu, Zhang, Shanfeng, Rao, Vikram, Mariserla, Sandeep, Chen, Ke, Naim, Iftekhar, Shahi, Shahrokh, Chen, Koert, Moiseev, Fedor, Walker, Trevor, Henrique Schechter Vera, Jain, Aashi, Han, Jay, Han, Feng, Ren, Xiaoqi, Yip, Cathy, Chen, Feiyang, Gupta, Nithi, Cer, Daniel, Shanbhogue, Madhuri

2025

Overview

In this report, we introduce Gemini Embedding, a state-of-the-art embedding model leveraging the power of Gemini, Google's most capable large language model. Capitalizing on Gemini's inherent multilingual and code understanding capabilities, Gemini Embedding produces highly generalizable embeddings for text spanning numerous languages and textual modalities. The representations generated by Gemini Embedding can be precomputed and applied to a variety of downstream tasks including classification, similarity, clustering, ranking, and retrieval. Evaluated on the Massive Multilingual Text Embedding Benchmark (MMTEB), which includes over one hundred tasks across 250+ languages, Gemini Embedding substantially outperforms prior state-of-the-art models, demonstrating considerable improvements in embedding quality. Achieving state-of-the-art performance across MMTEB's multilingual, English, and code benchmarks, our unified model demonstrates strong capabilities across a broad selection of tasks and surpasses specialized domain-specific models.

• Publisher: Cornell University Library, arXiv.org
• Subject: Benchmarks / Clustering / Embedding / Large language models