OpenAI Text Embeddings 3 Large
High-performance text embeddings for semantic search, RAG, and clustering
OpenAI's text-embedding-3-large model produces 3072-dimensional dense embeddings. Optimized for long-context text (up to 8192 tokens). API-based — no GPU required.
When to use:
- Semantic search over large document collections
- RAG retrieval pipelines (query and document embeddings)
- Text clustering and similarity grouping
- Anomaly detection in text data
Input: Text string Output: 3072-dimensional embedding vector
Inference Settings
Normalize Output (default: true) Normalize the embedding vector to unit length (L2 norm = 1).
- true: Use for cosine similarity — enables efficient dot-product similarity
- false: Keep raw magnitudes — needed if downstream models expect unnormalized vectors