The Import Tax Is Gone

What this post added

Introduced a significant price reduction for bulk import (free up to 1TB, then $0.25/GB) and increased the import capacity to 1TB or 100 million records per operation. Detailed the technical workflow of bulk import, contrasting it with the standard write path by explaining how it reads directly from object storage into the index builder for efficiency. Provided code examples for generating embeddings, writing Parquet files to object storage partitioned by namespace, and initiating/monitoring the import process. Clarified that bulk import is not compatible with indexes having schema definitions (full-text search, integrated embedding indexes).