Voyage AI review

Voyage AI bills per token and input pricing starts at $0.02. That is well below the €0.07 group median, and it places the brand high on entry price, ranked 11 of 71 on this measure. Because the catalogue is specialist, most spend goes on embeddings and reranking rather than chat tokens.

Yes, free tokens are included per account for the voyage embedding and rerank models, with the larger grant on the current voyage models and a smaller one on the older multilingual, finance, law and code models. That lets a prototype run end to end before you pay the $0.02 entry rate.

The catalogue in voyage 4 large, voyage 4 and voyage 4 lite covers general and domain embeddings, code, finance and law variants, multimodal picks, and a separate rerank family such as rerank-3. If your pipeline only needs retrieval vectors, that is a focused list rather than a wide grab bag.

The largest context window in the catalogue is 32,000, which is modest beside the 1,000,000 group median. On this measure the brand ranks 70 of 72 among compared APIs, so plan for chunking rather than very long single documents.

The certifications listed in SOC 2 and HIPAA are already in place, which helps when a security or privacy review blocks a project. That matters most for teams handling customer text, and it is one of the clearer reasons to pick a specialist embedding vendor.

Support runs through email, community and help centre, so documentation and email carry the load. There is no live chat or phone line in that list, which is worth checking against your own response time needs before you commit to a migration.

On this site's scoring the brand takes 4.5 overall, with cost contributing 8.6 and the offer part coming in at 0.4. In plain terms, pricing is the draw and the catalogue is the limit, so judge it on token volume rather than model count.