Valkey & Redis Pricing
How Valkey, our Redis-compatible key-value database, is priced per plan, what a plan fixes, and what is included.
Valkey, our managed Redis-compatible key-value database, is priced per plan.
| What | Rate | Detail |
|---|---|---|
| Valkey | from USD 3 / month | Priced by dataset size, on our own infrastructure. A persistent volume is included |
| Persistence | Included | Snapshotting to disk does not change the price |
What you get
Valkey is the Linux Foundation's BSD-licensed fork of Redis 7.2, compatible
with it at the protocol and command level. Your client library, your rediss://
connection string and your code work unchanged — there is nothing to port.
It runs on Cosmoner hardware, which is why it is priced the way it is: there is no third party between you and the machine.
Cosmoner previously resold managed Redis plans alongside this. That line is retired and no longer purchasable. Databases created on it keep running and keep being billed at their original price — nothing changes for them, and they can still be managed and deleted from the control panel as before.
What a plan fixes
A plan sets the dataset size it can hold. The plan price is the whole bill — commands, connections, and bandwidth are not metered on top of it.
Plans reserve CPU, stated in vCPU, and do not throttle. A single instance executes commands on one thread, so every tier reserves the same amount: tiers differ in how much data they hold, not in how fast they serve it.
There is no operations-per-second rating, deliberately. Nothing in the path enforces one, so quoting a number would be a promise rather than a limit.
High availability
Not offered. A failover needs something watching the primary and moving traffic to a replica when it goes, and nothing in the cluster does that yet — so it is not sold.
The plan list reports this per tier, and creating a database with replication requested is rejected rather than silently ignored.
Persistence
Persistence is included in every plan, and choosing a mode never changes the price. A persistent volume is provisioned whatever you pick, so the choice is purely about durability.
Selecting no persistence means we make no durability promise for that database — not that data is deliberately discarded. Treat it as a cache either way.
Regions
Valkey runs where we own capacity, which today is one region: Stockholm, Sweden. Should more regions be added, a plan will cost the same in each of them, so the choice will be about latency and data residency rather than price.
Capacity
Plans run on hardware we own, so a plan can be temporarily unavailable when the cluster is full. Creating one then fails before you are charged, and nothing is left behind — try a smaller plan, or contact us.
Deleting a database
Deleting stops the billing and credits the unused part of the period. The dataset is destroyed with it — export or snapshot first if you need the contents.