Repository Public
1# Gemini Embedding 2 Preview model23<br />45Our first multimodal embedding model, providing efficient numerical mapping of6text, images, video, audio, and PDFs into a single unified embedding space. The7Gemini Embedding 2 model is best for cross-modal semantic search, document8retrieval, and recommendation systems that require fast, scalable similarity9calculations across large multimodal datasets.1011## Documentation1213Visit the [Embeddings](https://ai.google.dev/gemini-api/docs/embeddings) page for full coverage14of features and capabilities.1516## gemini-embedding-2-preview1718| Property | Description |19|---|---|20| Model code | **Gemini API** `gemini-embedding-2-preview` |21| Supported data types | **Input** Text, image, video, audio, PDF **Output** Text embeddings |22| Token limits^[\[\*\]](https://ai.google.dev/gemini-api/docs/tokens)^ | **Input token limit** 8,192 **Output dimension size** Flexible, supports: 128 - 3072, Recommended: 768, 1536, 3072 |23| Versions | Read the [model version patterns](https://ai.google.dev/gemini-api/docs/models/gemini#model-versions) for more details. - Preview: `gemini-embedding-2-preview` |24| Latest update | March 2026 |