Voyage voyage-3 logo

Voyage voyage-3

Free

Top MTEB retrieval scores

FreeFree tier
Inputs: text
Type
Open Source
Company
Voyage AI

About Voyage voyage-3

Voyage voyage-3 is a text embedding model from Voyage AI, optimized for general-purpose and multilingual retrieval tasks. It supports a context length of 32,000 tokens and produces 1024-dimensional embeddings. As a previous-generation model, it remains accessible through the Voyage API but does not include a free tier, with usage charged per token. It is part of a broader family of models designed for diverse retrieval and RAG applications.

Key Features

32,000 token context length
1024-dimensional embeddings
Optimized for general-purpose retrieval
Optimized for multilingual retrieval
Accessible through Voyage API
Previous-generation model (still available)

Pros & Cons

Pros
  • Strong retrieval quality for general and multilingual content
  • Long context length of 32,000 tokens
  • Part of a well-established embedding model ecosystem
  • Compatible with Voyage API infrastructure
Cons
  • Previous-generation model; newer models offer better quality and efficiency
  • No free tier available (unlike newer models)
  • Pricing per token may be higher than some alternatives

Best For

General-purpose semantic searchMultilingual document retrievalRetrieval-augmented generation (RAG) pipelinesCross-lingual information retrievalEmbedding-based text clustering and classification