PayloadIndexParams

The parameters Qdrant accepts beside the index type, one shape per type.

PayloadSchemaType names the kind of index; this names how it is built. Two of the choices decide whether a query works at all rather than how fast it is: a matchPhrase filter matches nothing unless the text index was created with Text.phraseMatching, and an index built without Keyword.isTenant leaves a multi-tenant collection's points scattered across the disk rather than grouped per tenant. The rest are cost: onDisk is the difference between an index that has to fit in RAM and one that does not.

Build one with the DSL rather than by hand:

qdrant.createPayloadIndex("docs", "body") {
    text { tokenizer = Tokenizer.WORD; phraseMatching = true }
}

Qdrant takes two further text options this does not model, stopword sets and stemming, because both are sub-objects with their own vocabulary rather than a flag. They are the reason this type is a sealed hierarchy that can grow a field without breaking a caller.

Inheritors