Text-Embedding-V4 is a text embedding model designed to convert text into high-dimensional vector representations. It captures semantic meaning for downstream tasks such as retrieval, clustering, and classification.
Key Features
- High semantic fidelity: Produces embeddings that preserve nuanced meaning across sentences and paragraphs.
- Multilingual support: Handles multiple languages within a shared vector space.
- Scalable dimension: Offers configurable output dimensions to balance accuracy and storage.
- Efficient inference: Optimized for low-latency batch processing.
- Robust to domain shift: Maintains performance on specialized or technical text.
Use Cases
- Semantic search and retrieval-augmented generation (RAG)
- Document clustering and topic discovery
- Duplicate detection and deduplication
- Recommendation systems based on text similarity
- Zero-shot classification via embedding comparison
