text-embedding-3-large
text-embedding-3-large converts text into vectors for matching, retrieval and grouping. It supports English and other languages and returns numerical representations rather than chat answers.
Vector dimensions can be reduced
The default vector has 3,072 dimensions. The dimensions parameter can request a shorter representation. The model page lists $0.13 per million input tokens; storage and vector-search costs belong to the separate retrieval system.text-embedding-3-large official model documentationOpenAI embedding dimensions and retrieval guide
Choose dimensions with a retrieval comparison
Practical guidance. Use representative queries and known relevant documents to compare shorter vectors. Measure missed results as well as storage size. Keep query and document embeddings on the same model and dimension settings when building an index.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | API |
| Model identifiers | text-embedding-3-large |
| Input price | 0.13 USD per million tokens Published embedding input-token rate text-embedding-3-large official model documentation |
| Output | Numerical text vectorstext-embedding-3-large official model documentation |
| API routes | Batch (v1/batch); Embeddings (v1/embeddings)text-embedding-3-large official model documentation |
Related models
- text-embedding-3-small
OpenAI · OpenAI embeddings
text-embedding-3-small provides text vectors for search and relatedness tasks.
Launch date not confirmedAPI - text-embedding-ada-002
OpenAI · OpenAI embeddings
text-embedding-ada-002 is an older text embedding model that remains in the public catalog.
Launch date not confirmedAPI