About Voyage AI
Voyage AI provides embedding models and rerankers for search and retrieval. It is part of MongoDB and aimed at retrieval-augmented generation and enterprise search.
What it does
Embedding models turn text, code or documents into vectors, and rerankers reorder search results by relevance. Developers call the models through an API and pair them with the vector database of their choice. Smaller vectors can reduce the cost of storage and search for large collections.
Key features
- Embedding models for text and code
- Rerankers to improve search order
- Context length of up to 32K tokens
- Available through major cloud providers
- Custom and in-VPC deployment options
Who it's for
- Teams building retrieval pipelines for AI apps
- Companies with large document collections
- Developers who want accurate search without a large model
Pricing and good to know
Pricing is consumption-based, so you pay for what your application processes. Custom deployments and model licensing are arranged with sales. Check the pricing page for current model rates, and test retrieval quality on your own documents before choosing a model.
Highlights
- Create embeddings for text and code
- Rerank search results by relevance
- Handle up to 32K tokens per input
- Use through major cloud providers
- Deploy in your own VPC
