Amazon Titan Text V2
Awaiting activationBuild English enterprise search with Amazon’s embedding model on Bedrock.
What it does
Amazon Titan Text Embeddings V2 is an Amazon Bedrock model optimized for text retrieval. It maps a document, paragraph or sentence into a vector so a search index can find relevant information or group related content. AWS offers on-demand and provisioned-throughput inference.
The model returns 1,024-dimensional vectors by default and supports 512 or 256 dimensions. AWS documents an 8,192-token input window. This service returns normalized float vectors, accepts one text per request and enforces an 8,000-byte limit.
Language coverage
Amazon says Titan Text V2 is optimized for English. Its preview support for other languages is excluded from this directory’s named-language recommendations.
Official language coverage ↗Where it helps
Enterprise search
Find useful English passages in manuals and business records.
Customer support
Match a question with a relevant support answer.
Document organization
Group similar English content for browsing and discovery.
When to choose it
Consider it for English-first retrieval and an AWS-based model option. Compare its smaller vector sizes when index storage matters.
Awaiting Bedrock activation, regional access and verified account rates. AWS describes other languages as preview support and cautions that cross-language retrieval can be weaker; do not treat it as equally capable in every language.
Official sources
Paraphrased from the provider’s documentation and original model information. Selection advice is our editorial guidance.
Amazon Web Services · Titan Text Embeddings model guide ↗Reviewed October 4, 2026