landing_page/qdrant-landing/content/quantization/configuration.md

label, title, tables, banner, sitemapExclude

label	title	tables	banner	sitemapExclude
CONFIGURATION	Choose Your Method: The Comparison Matrix	id featureCellWidth cols features quantization-configuration 13rem id name highlight bold scalar Scalar false false id name highlight bold turboQuant TurboQuant false false id name highlight bold binary Binary false false id name highlight bold product Product false false name scalar turboQuant binary product Memory Cut 4x 8x to 32x Up to 32x Up to 64x name scalar turboQuant binary product Typical Recall (with rescoring) Usually within 1% Comparable <br>to scalar at double the compression High on centered, high-dim embeddings Lower, tune carefully name scalar turboQuant binary product Speed Faster Fast Fastest (up to 40x) Slower name scalar turboQuant binary product Best for Safe default, <br>any dimensionality Strong default, <br>no dataset training Models with 1024+ dimensions When memory <br>is the only priority	content link Scalar, product, and binary quantization each make a different tradeoff across memory savings, recall, and query speed. url text /documentation/manage-data/quantization/ Compare Quantization Methods	true