Text embedding
Nomic Atlas natively supports the following text embeddings models:
| Model | Description | Context Length | Dimension | MTEB |
|---|---|---|---|---|
| nomic-embed-text-v1 | Nomic Embed specialized for retrieval, similarity, clustering and classification. | 8192 | 768 | 62.39 |
| nomic-embed-text-v1.5 | Nomic Embed with support for variable embedding size and specialized for retrieval, similarity, clustering and classification. Recommended output sizes are 768, 512, 256, 128 and 64. | 8192 | 768 | 62.28 |
| 512 | 61.96 | |||
| 256 | 61.04 | |||
| 128 | 59.34 | |||
| 64 | 56.10 | |||
| gte-multilingual-base | A multilingual text embedding that supports over 70 languages. | 8192 | 768 | 61.40 |
Visit the API Reference for use details.
*Model quality degrades with decreased output dimensionality. Learn more.
Embedding task types
There are four task types for Nomic Embed:
- search_query: A query for retrieval.
- search_document: A document for retrieval, or a query for similarity search.
- classification: Used for classification tasks.
- clustering: Results in very high linear separability.