text-embedding-3-large
OpenAIEmbeddingProprietaryOpenAI's most capable embedding model for converting text into vectors. Used for semantic search, clustering, and RAG.
Abilities
Use Cases
Available in Tools
Availability
How to Use
Pros
- High-quality embeddings — top scores on MTEB benchmark for retrieval tasks
- Configurable dimensions (256–3072) — trade quality for storage/speed
- Simple API — just send text, receive vectors, no ML expertise needed
- Wide language support for multilingual embedding applications
Cons
- Closed source — cannot self-host, all data must be sent to OpenAI
- Requires API calls — no offline use, adds latency to every request
- Cost adds up with large document collections ($0.13/1M tokens)
- Cannot fine-tune — stuck with generic embeddings
- Open-source alternatives (BGE, GTE) are approaching comparable quality for free
What to Use It For
Perfect For
Top MTEB benchmark scores for retrieval — best general-purpose embedding quality available via API
High-quality vector representations ensure relevant document chunks are retrieved for generation
Configurable dimensions (256-3072) allow trading quality for speed in clustering and classification tasks
Good For
High-quality semantic similarity makes it effective at finding near-duplicate content
Vector embeddings enable content-based recommendations by finding semantically similar items
Not Recommended
All data must be sent to OpenAI servers — no offline or on-premise option for data that cannot leave your network
Try instead: BGE, GTE
Cannot be fine-tuned — stuck with generic embeddings that may miss domain-specific nuances
Try instead: Cohere Embed
Do Not Use For
Embedding model only — converts text to vectors, cannot generate any text or responses
Try instead: GPT-4o
API call overhead adds network latency to every embedding request — too slow for sub-millisecond requirements
Try instead: local embedding model