We no longer provide the Optimized for Vectors option in the UI, it should only be visible in the API.
The short answer is, unless you have very specific needs then you want General Purpose.
For most search use cases, including text search, as well as most vector and hybrid search use cases, General Purpose is the best fit.
Vector Optimized is weighted to provide more RAM relative to CPU and Disk for use cases that rely on dense vectors with HNSW (which is a memory intensive algorithm).
By default, Serverless uses DiskBBQ instead of HNSW which does not that same dependency on RAM.
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.