Back to Models
Google Deep MindConditional Open
EmbeddingGemma - 300M
EmbeddingGemma - 300M is an embedding model developed by Google DeepMind. It enables efficient vector representations and is suitable for semantic search and classification of text data.
Parameters
3.0B
Context Window
2K
License
Gemma License
Release Date
2025-09-05
API Pricing
API pricing for this model is not yet available
Strengths
- ・Developed by Google DeepMind
- ・Lightweight model size
- ・Efficient vector transformation
Weaknesses
- ・Short context length of 2K
- ・Limited parameter scale
- ・Specialized model, not general-purpose
Use Cases
- ・Implementing semantic search
- ・Assessing document similarity
- ・Text clustering
Deep Analysis
Parameters
308M
MTEB Multilingual
61.15
SOTA for sub-500M
MTEB English
69.67
MTEB Code
68.76
Release Date
September 24, 2025
Strengths
- ・SOTA on MTEB for sub-500M models
- ・Comparable to models double its size
- ・Robust to quantization
- ・Ideal for on-device
Weaknesses
- ・Embedding-only
- ・Google usage license
- ・Text-only
Competitor Comparison
| Model |
|---|
| NV-Embed-v2 |
| GritLM-7B |
EmbeddingGemma is Google DeepMind's 308M parameter embedding model. SOTA on MTEB for sub-500M models.
Sources
Analysis generated: 2026-05-24