text-embedding-3-large
Gold standard dense embedding model featuring Matryoshka learning, allowing developers to truncate dimensions without retraining.
Best for: Enterprise RAG retrieval pipelines, semantic deduplication, and vector search clusters.
Specifications
- Provider: OpenAI
- Release date: January 2024
- Context window: 8,191 tokens
- Modalities: Embedding
- License: Proprietary Commercial API
- Pricing: $0.13 / MTok
- API available: Yes
Capabilities
- 3,072 dimensional dense vectors
- Matryoshka representation learning
- Flexible dimensionality shortening
- High semantic clustering