BetterDB vs Redis Iris
Redis Iris is Redis's enterprise context engine. BetterDB is the open-core alternative: the same memory, caching, and retrieval primitives on the open-source Valkey fork, with no enterprise contract and no lock-in.
What Redis Iris is
Redis Iris (the Context Engine, launched May 2026) is Redis’s enterprise context and memory platform, built from five tools: Context Retriever, Agent Memory, Data Integration (RDI), LangCache, and Redis Search, with SSD tiering (Flex) to hold larger contexts and longer memories more cheaply. It is the most direct strategic parallel to BetterDB — but it is a proprietary offering delivered through Redis Enterprise and Redis Cloud, with the Context Retriever and Agent Memory components in preview.
BetterDB vs Redis Iris, side by side
| BetterDB | Redis Iris | |
|---|---|---|
| What it is | Open context layer: memory + cache + retrieval | Enterprise context engine: retriever, memory, RDI, LangCache, search |
| License | Open core on the Valkey fork | Proprietary (Redis Enterprise / Redis Cloud) |
| Datastore | A Valkey anywhere — self-hosted, your managed cloud, or managed by us | Redis Enterprise / Cloud, with Flex SSD tiering |
| Lock-in | None — your infrastructure, open core | Redis Enterprise / Cloud and its licensing |
| Semantic LLM cache | Yes — bundled, exact + semantic, multi-tier | Redis semantic cache, configured separately |
| Languages | TypeScript + Python parity | Redis client ecosystem |
| Observability | OpenTelemetry + Prometheus at every layer | Redis Enterprise tooling |
| Availability | Generally available, self-host today | Agent components in preview |
Why teams pick BetterDB over Redis Iris
Open, not enterprise-gated
The same three primitives — memory, cache, retrieval — without an enterprise license or a sales contract.
Runs on Valkey you already operate
Build on the open-source fork in your own infrastructure instead of committing to Redis Enterprise or Redis Cloud.
No vendor lock-in
Avoid being tied to Redis's licensing direction. Valkey is the open, community-governed fork.
Observable by default
OpenTelemetry and Prometheus across memory, cache, and retrieval, with bundled cost tracking.
One SDK, TypeScript and Python
A single typed SDK with full language parity, not a stack of enterprise services to wire together.
Where Redis Iris is stronger
No tool wins everywhere. Here is where Redis Iris is the better choice.
Enterprise scale and tiering
Flex SSD tiering targets very large contexts and long memories at enterprise scale.
First-party data integration
RDI keeps context fresh from relational databases and warehouses out of the box.
Brand, support, and compliance
Established enterprise support, SLAs, and compliance posture under the Redis name.
A note on benchmarks: published memory-accuracy numbers across this category are rarely comparable. LongMemEval has small (S) and large (M) splits, and scores swing with the reader model, judge model, embedding model, and k. Our ~93% figure is recall (with hybrid rerank) on the larger LongMemEval-M split, which is a different metric and dataset from the QA-accuracy or J-scores vendors usually headline. We do not publish a head-to-head accuracy number against this product, because no apples-to-apples run exists.
Build your context layer on Valkey
Install the SDK and get agent memory, semantic caching, and retrieval in one library. Self-host on a Valkey you already run — or let us provision a managed Valkey with the search module, no setup required.