Redis vs Valkey: Licences, Compatibility and Which to Pick in 2026
Redis vs Valkey explained: the 2024 licence change, the Linux Foundation fork, Redis 8's AGPLv3 option, compatibility, managed services and how to choose.
Valkey is the Linux Foundation fork of Redis 7.2.4, created in March 2024 after Redis moved away from the BSD licence. Both are fast in-memory key-value stores that speak the same protocol, so for a typical app (caching, sessions, queues, rate limits) either works and most client code runs unchanged. The real differences are licensing, which managed services offer which engine, and the newer features each project has added since the split.
Versions and licences below were checked against redis.io, valkey.io and the two GitHub release pages in November 2026.
What happened: the short history
- Up to Redis 7.2: Redis was BSD-3-Clause, a permissive open source licence.
- March 20, 2024: Redis Ltd. announced that Redis 7.4 and later would be dual-licensed under the Redis Source Available License v2 (RSALv2) and the Server Side Public License v1 (SSPLv1). Neither is an OSI-approved open source licence. Redis's own licence page says so for both.
- March 28, 2024: The Linux Foundation launched Valkey, forked from Redis 7.2.4 and kept under BSD-3-Clause. It launched with backing from AWS, Google, Oracle, Ericsson and Snap, and with maintainers from companies including Alibaba, Huawei and Tencent.
- November 2024: Salvatore Sanfilippo (antirez), Redis's original author, rejoined Redis.
- May 2025: Redis 8 added AGPLv3 as a third licence option. From Redis 8 onwards you can choose RSALv2, SSPLv1 or AGPLv3. AGPLv3 is OSI-approved, so Redis is open source again, though copyleft rather than permissive.
Redis's licence page sums it up by version:
| Redis version | Licence |
|---|---|
| 7.2 and earlier | BSD-3-Clause |
| 7.4 | RSALv2 or SSPLv1 |
| 8.0 and later | RSALv2, SSPLv1 or AGPLv3 (your choice) |
Valkey has stayed BSD-3-Clause throughout.
Side by side
| Redis (Open Source) | Valkey | |
|---|---|---|
| Governance | Redis Ltd. | Linux Foundation project |
| Licence | RSALv2 / SSPLv1 / AGPLv3 (8.0+) | BSD-3-Clause |
| Latest stable (Nov 2026) | 8.10.3, with 8.12 in release candidates | 9.1.3, with 9.2 in release candidates |
| Built-in extras | JSON, Query Engine (search), time series, probabilistic types, vector sets | Core server; JSON, Bloom, search and LDAP via modules (valkey-bundle) |
| Docker image | redis (Docker Official Image) |
valkey/valkey, valkey/valkey-bundle |
| AWS ElastiCache / MemoryDB | Redis OSS engine still offered | Offered, priced lower than Redis OSS |
| Google Memorystore | Memorystore for Redis | Memorystore for Valkey (GA since April 2025) |
Version numbers have diverged, so don't compare them directly. Valkey 9.1 is not "newer" than Redis 8.10. They are separate lines that started from the same 7.2 code.
What the licences mean for you
For most people building an app, neither licence affects you in practice. You can run Redis 8 or Valkey as the cache behind your own SaaS without any obligation, because you're using the server, not selling it.
Where it matters:
- RSALv2 forbids offering Redis as a managed service to third parties. If you're building a hosting product that sells "Redis as a service", that's the clause you'd hit.
- AGPLv3 requires that if you modify the Redis server and let users interact with it over a network, you offer them your modified source. Running an unmodified Redis behind your app doesn't trigger that. Many companies still have a blanket "no AGPL" policy, so check with whoever owns that list.
- BSD-3-Clause (Valkey) asks only that you keep the copyright notice. Nothing to think about.
If your legal team, a client contract or a Linux distribution's policy prefers permissive licences, Valkey is the simple answer. For a deeper primer on the licence families, see open source licenses explained.
Compatibility: can you switch?
Valkey's migration guide says it is compatible with Redis OSS 2.x through 7.2.x: same protocol, same configuration, same RDB and AOF file formats. Moving from Redis 7.2 or earlier to Valkey is effectively an upgrade. You can copy the RDB snapshot across, set up replication, or use MIGRATE for specific keys.
The catch is the other direction and newer versions. Per the same guide, Redis 7.4 and later write data files that Valkey cannot load. If you're on Redis 7.4 or 8.x today and want to move, you can't just drop the dump file into Valkey. You'd need to re-populate the data (fine for a cache) or copy keys with a tool that reads from one server and writes to the other.
At the command level, core data types and commands (strings, hashes, lists, sets, sorted sets, streams, pub/sub, Lua scripting, EXPIRE, INCR and so on) behave the same. Divergence shows up at the edges:
- Redis 8 bundles what used to be Redis Stack:
JSON.*commands, the Query Engine (FT.*), time series, Bloom/cuckoo filters, plus the new vector sets type. - Valkey keeps the core lean and ships extra capabilities as modules. The
valkey/valkey-bundleimage preloads valkey-json, valkey-bloom, valkey-search and valkey-ldap. These are similar in purpose, but don't assume every Redis Stack command and option exists one-for-one.
If your app uses only plain caching, sessions and queues, you will almost certainly not notice which one you're on. If it relies on FT.SEARCH or JSON.SET, test against the exact server you'll deploy.
Client libraries
The existing Redis clients (redis-py, ioredis, node-redis, go-redis, Jedis, Lettuce) speak the same wire protocol, so they generally work with Valkey. Valkey's own clients page lists:
- Node.js: Valkey GLIDE (
@valkey/valkey-glide) andiovalkey, a fork of ioredis - Python:
valkey-glideandvalkey(valkey-py, a fork of redis-py) - Go: Valkey GLIDE for Go and
valkey-go - Java: Valkey GLIDE,
valkey-java(Jedis-derived) and Redisson
GLIDE is a multi-language client with a shared Rust core, aimed at cluster reliability. You don't have to adopt it. Switching servers usually means changing the connection URL and nothing else.
# Try either locally
docker run -d --name redis -p 6379:6379 redis:8
docker run -d --name valkey -p 6380:6379 valkey/valkey:9.1
# Same CLI commands work against both
redis-cli -p 6379 SET greeting hello
redis-cli -p 6380 SET greeting hello
Managed services
This is where Valkey has pulled ahead commercially:
- AWS offers Valkey on both ElastiCache and MemoryDB. The ElastiCache pricing page lists Valkey at 20% lower than Redis OSS for node-based clusters and 33% lower on Serverless. MemoryDB for Valkey launched at 30% below MemoryDB for Redis OSS. The Redis OSS engine is still available, with Extended Support for older major versions.
- Google Cloud has Memorystore for Valkey, generally available since April 2025, alongside Memorystore for Redis.
- Redis Cloud, run by Redis Ltd., is the place to get Redis 8 features fully managed, along with commercial support from the company that writes it.
Many smaller hosts and PaaS providers now offer one or both. Check what your host runs before you plan around a specific module.
How to pick for a small app
- You're starting fresh and want the simplest licence: Valkey. BSD, foundation governance, cheaper on AWS.
- You're on AWS ElastiCache or MemoryDB: Valkey, for the price difference alone, unless you depend on something only the Redis engine has.
- You want built-in JSON documents, full-text search or vector search inside your cache: Redis 8 has these in the core server with long-standing docs. Valkey's modules cover similar ground but are younger.
- You already run Redis 7.2 or older: you can stay put for now or move to Valkey with a straight data copy. Both are maintained.
- You already run Redis 7.4 or 8.x and it works: there's no urgent reason to move. If your company is fine with AGPLv3 or you never modify the server, the licence probably doesn't bite.
- Your company forbids AGPL and SSPL: Valkey.
For most small apps the honest answer is that the choice matters less than whether you need either at all. A single Postgres database can handle caching, queues and locks until traffic says otherwise. See when you actually need Redis, and caching explained for the patterns either server supports. If you do add one, BullMQ and distributed locks are the most common next uses.
EasySpawn gives each server a managed Redis instance next to your app, with a ready-made REDIS_URL your code can connect to. See how it works or join the waitlist.
Related: Open Source Licences Explained for People Building Apps · Redis: When a Small App Actually Needs It (and When Postgres Is Enough) · What Is Caching? A Beginner's Guide to Making Apps Faster · BullMQ Tutorial: Background Jobs in Node.js With Redis
Keep reading
Supabase vs Convex: Postgres Backend or Reactive TypeScript Backend?
Supabase vs Convex: Postgres with SQL and RLS against a reactive document database with TypeScript functions. Realtime, auth, self-hosting, lock-in, pricing.
Drizzle vs Prisma: Which TypeScript ORM Should You Use in 2026?
Drizzle and Prisma are the two leading TypeScript ORMs. Drizzle is SQL-shaped, lightweight and schema-in-TypeScript; Prisma has its own schema language, a higher-level query API and, since Prisma 7, a Rust-free client. How they compare on schema, queries, migrations, serverless, performance and AI-assisted coding.