Supercharging
Search
and
Retrieval
for Unstructured
Data
Best-in-class embedding models and rerankers
Embeddings and Rerankers Drive RAG Retrieval and Response Quality
Unstructured data
Embedding model
Vector DB
Reranker
Relevant files
LLM
Factual responses with lower costs
A Spectrum of Models for Your Target Use Cases
General-purpose models
Ready for any purpose and language out-of-the-box.
Domain-specific models
Highly optimized for industry-specific data, like finance, legal, and code.
Company-specific models
Fine-tuned librarians for your company’s unique data and lingo.
Powered by Cutting-Edge AI Research and Engineering
High accuracy
Retrieving the most relevant contextual information
Low dimensionality
3x-8x shorter vectors ⇒ cheaper vector search and storage
Low latency
4x smaller model and faster inference with superior accuracy
Cost efficient
2x cheaper inference with superior accuracy
Long-context
Longest commercial context length available (32K tokens)
Modularity
Plug-and-play with any vectorDB and LLM
Trusted by Industry Leaders
Continue allows engineers to converse with entire codebases and beyond, making Voyage AI’s code embedding model a natural choice for our use case. The combined performance of voyage-code-2 plus rerank-lite-1 in our benchmarks made working with Voyage AI a no-brainer, and they have proven to offer unmatched accuracy and stability.
Harvey has partnered with Voyage AI to create custom legal embeddings tailored to Harvey’s use cases. We chose Voyage because their voyage-law-2 legal embedding model has already demonstrated best-in-class performance on public legal retrieval benchmarks. Fine-tuning the model with Harvey’s specific datasets reduces the irrelevant document rate by 25% and decreases the dimensionality of the embeddings and the vectorDB costs by threefold.
Excellent code retrieval accuracy is crucial for Replit Agent, as it allows us to always put in context the most relevant files depending on the action type. The Voyage AI team has deep expertise developing embedding models for domain-specific retrieval, and the state-of-the-art performance attained by voyage-3 will give us the best overall performance for our code retrieval tasks.
At Snowflake, we’re dedicated to delivering exceptional customer experiences. That’s why we’ve partnered with Voyage AI, whose embedding models have consistently demonstrated unmatched quality in real-world applications. By integrating Voyage AI’s superior technology with the security, efficiency, and governance of Snowflake’s Data Cloud, we’re unlocking unparalleled Retrieval-Augmented Generation (RAG) experiences for our customers. With Voyage AI’s best-in-class models now available as native features within our platform, we’re setting a new standard for AI-driven innovation and customer satisfaction.
End user experience is extremely important to us at Kapa.ai, and a big part of that is making sure our AI assistants are grounded with reliable sources of knowledge. Voyage AI’s powerful reranker made a big difference for us - not only did it perform the best, but it was also easy to integrate into our existing stack, unlocking significant performance gains and enabling highly accurate retrieval-augmented generation at a low cost.
Our collaboration with Voyage AI to launch Noxtua Voyage Embed marks a significant step in advancing specialized, high-quality legal AI solutions, demonstrating a 1.7x better retrieval accuracy than OpenAI’s latest text-embedding-3-large embedding model on legal benchmarks. This cross-border collaboration has allowed us to further develop Noxtua, Europe’s first independent legal AI, to better assist lawyers in their everyday tasks.
We use cookies and similar technologies to deliver, maintain, improve our services and for security purposes.
Please review our Privacy Policy for details. Click ‘Accept all’ to let Voyage AI and partners use cookies for these purposes, or click ‘Reject non-essential’ to decline all cookies except those that are strictly necessary.