EmbeddingGemma

Parameters

308M

Published

Sep 5, 2025

EmbeddingGemma is an AI model developed by Google DeepMind (United States), first published in September 2025. It works in the language domain, on tasks such as semantic embedding.

Epoch AI has no training-compute estimate for this model. The model has 308,000,000 parameters. It was trained on roughly 320B datapoints. Training ran on Google TPU v5e.

Access: Open weights (restricted use). Its weights are openly available. It is built on top of Gemma 3 270M. Epoch AI rates the confidence of this record as confident.

Full record

Organization: Google DeepMind
Country of organization: United States
Domain: Language
Task: Semantic embedding
Parameters: 308,000,000
Dataset size: 320B
Training hardware: Google TPU v5e
Model accessibility: Open weights (restricted use)
Open weights: Yes
Base model: Gemma 3 270M
Epoch confidence: Confident