Vector Sets have been integrated into Redis as a new core data type that allows users to handle vectors for similarity searches, similar to Sorted Sets but with vectors as scores. The implementation emphasizes a simple API, efficient deletion, threading for performance, and innovative features like quantization and JSON filtering. This marks a significant addition to Redis, enhancing its capabilities for managing complex data structures.

Redis creator Salvatore Sanfilippo has returned and introduced a new data type called vector sets, designed for storing and querying high-dimensional embeddings for AI workloads. This development is part of Redis's evolution beyond caching, and includes new features like LangCache, a semantic caching service aimed at optimizing interactions with large language models.