LLM inference

Valkey has released RC1 for version 9.2, its largest launch to date, featuring forkless replication that dramatically reduces memory overhead and a new hierarchical data structure designed for LLM inference routing. The project is also advancing synchronous replication, a move that would allow Valkey to serve as a durable data store rather than purely a cache. For organizations evaluating open source in-memory infrastructure, this release reframes what Valkey can credibly do.