# Valkey 9.2 adds shared caching for AI inference

> The open-source datastore adds distributed KV caching and changes aimed at cutting backup memory, replication traffic and large sorted-set overhead.

Oossa · 2026-11-19 · https://oossa.com/en/valkey-9-2-adds-shared-caching-for-ai-inference

Valkey announced general availability of version 9.2, alongside an open-source distributed key-value cache for AI inference. It lets inference engines reuse computed prompt and conversation data across a GPU fleet; teams using self-managed open-weight models can connect it through LMCache with vLLM or SGLang.

Valkey says the cache can cut inference costs by up to 40% and time-to-first-token by up to 80%. The release also adds forkless snapshots to avoid backup-related memory spikes, and replication compression that the company says reduces traffic by roughly 50% to 65%.

## The facts

- The distributed cache connects through LMCache to engines including vLLM and SGLang.
- Valkey says replication compression reduces ongoing traffic by roughly 50% to 65%.

## Why it matters

Teams running open-weight models on their own GPU fleets can try reusing inference work through LMCache, though the advertised savings are company claims, not a guarantee for every workload.

## Sources & references

1. [Valkey 9.2 Lowers Cost of AI Workloads On Existing Infrastructure Alongside New Distributed KV Caching Capability](https://www.prnewswire.com/news-releases/valkey-9-2-lowers-cost-of-ai-workloads-on-existing-infrastructure-alongside-new-distributed-kv-caching-capability-302945903.html) – PR Newswire RSS, 2026-11-19

Last updated: 2026-11-19
