Hobby
€8/mo
Always-on small app. Side projects, internal tools, demos.
Fixed price · per replicaApp hosting, managed data, and isolated sandboxes. All on one prepaid balance. Top up, then run. Plan prices, usage rates, and resource limits are listed below, so you can work out the cost before you deploy.
Choose a fixed price for sustained load or measured usage for quieter workloads. Set the size and billing model separately for each process type.
Start small. Add capacity as your app grows.
€8/mo
Always-on small app. Side projects, internal tools, demos.
Fixed price · per replica€15/mo
1 vCPU, 2 GiB RAM. Real production traffic at startup scale.
Fixed price · per replica€30/mo
2 vCPU, 4 GiB RAM. Headroom for growing apps.
Fixed price · per replica€60/mo
4 vCPU, 8 GiB RAM. Heavier web workloads or memory-bound workers.
Fixed price · per replica€120/mo
8 vCPU, 15 GiB RAM. Big in-memory work, image processing.
Fixed price · per replica€240/mo
16 vCPU, 30 GiB RAM. Single-replica heavy lifters.
Fixed price · per replicaStandard is priced for sustained load. Your chosen size sets the capacity available to each replica.
Your process stays running and reachable.
≤ €40/mo
Up to 1 vCPU and 2 GiB RAM. For quiet APIs, workers, staging, and review apps.
Monthly ceiling · per replica≤ €80/mo
Up to 2 vCPU and 4 GiB RAM, billed from measured use.
Monthly ceiling · per replica≤ €160/mo
Up to 4 vCPU and 8 GiB RAM, billed from measured use.
Monthly ceiling · per replica≤ €310/mo
Up to 8 vCPU and 15 GiB RAM, billed from measured use.
Monthly ceiling · per replica≤ €620/mo
Up to 16 vCPU and 30 GiB RAM, billed from measured use.
Monthly ceiling · per replicaEvery minute, Watasu meters actual CPU time and memory working set across your replicas.
The minimum is 0.01 vCPU and 0.01 GiB per running replica: €0.30 per replica-month. Scaling to zero stops compute charges.
0.05 vCPU + 0.2 GiB on average
Dynamic 1× can burst to 1 vCPU and 2 GiB, but can never bill above €40 per replica-month.
Multi-DC placement is automatic once a process has 2+ replicas, at no extra charge.
Run agent evals and production tasks entirely in the EU. Your team plan sets the limits; runtime and retained storage are billed separately.
Base price, before runtime and storage usage.
For short runs and getting started
Free base plan
Runtime and retained storage billed separately.
For longer runs and more parallel work
€150/mo base price
Runtime and retained storage billed separately.
For teams with custom requirements
Custom base price
Per-team overrides for disk, runtime, concurrency, CPU, memory, and create rate.
Runtime is billed per second for the vCPU and memory you request. Retained storage is prorated monthly.
Stopped sandboxes keep their disk for instant resume and pay only for that storage. Read the sandbox billing docs
Choose a service to explore every plan, resource allocation, and the command to attach it.
The default for transactional application data. Async followers from Standard 1, sync-standby HA on Premium. Hobby uses manual backups; scheduled backups kick in from Standard up. Restore-and-promote workflow on every tier.
Single-node, network-backed. Manual backups. For prototypes and review apps.
Local-NVMe production tier. Async followers from Standard 1. Scheduled backups by default.
Sync-standby HA + async followers. One stable write endpoint, automatic failover.
watasu addons:create postgresql --app my-app
Read the docs
In-memory data store for caching, sessions, rate limiting, and ephemeral coordination. Speaks the Redis protocol. Every Premium plan ships with AOF persistence, a managed HA standby, and automatic failover — pick a size, not a topology.
Single-node, non-persistent. For prototypes, review apps, and throwaway caches.
Production tier at every size — AOF persistence, managed HA standby, automatic failover. Scale by data set and connections.
watasu addons:create valkey --app my-app
Read the docs
For event data, time series, and dashboards over millions or billions of rows. Standard runs replicated across two zones; Premium adds multi-shard scale-out.
Single node, network-backed, non-production. Manual backups only.
Replicated production — 1 shard × 2 replicas across two zones, local NVMe, scheduled backups.
Multi-shard with 2 replicas per shard. Storage is listed per replica; replicas store copies of the same shard. Distributed tables, scheduled backups, multi-zone placement.
watasu addons:create clickhouse --app my-app
Read the docs
Native Qdrant REST and gRPC. Plans are defined per peer; replica copies consume physical disk. Daily backups on Standard and Premium. Premium spans three EU locations.
Single peer, network SSD. Manual capture.
Three peers, one location, local NVMe.
Peers distributed across three EU locations, local NVMe.
Durable, ordered, replayable event streams. Hobby is single-broker with throughput quotas; Standard and Premium are 3-broker (Standard) or 8-broker (Premium) replicated clusters.
Single broker with throughput quotas. Non-production, for development and small staging.
3-broker replicated cluster, no quotas. Local NVMe, RF=3, rack-aware placement.
8-broker cluster on large or XL pods, no quotas, 6-week retention.
watasu addons:create redpanda --app my-app
Read the docs
Per-app S3-compatible bucket with scoped credentials. Generous bundled egress. Premium tiers are geo-replicated across three zones.
Single-zone S3-compatible storage. For prototypes, asset hosting, small uploads.
Single-zone, production-grade scale and egress.
Geo-replicated across three zones (primary + two replicas).
watasu addons:create object-storage --app my-app
Read the docs
A persistent per-app Grafana workspace — dashboards, Explore, and alerts. Auto-attached the first time you add Logs, Metrics, or Traces, and stays free regardless of which observability add-ons you keep.
One workspace per app. Persistent dashboards, alert rules, and saved Explore views.
Auto-attached with any Logs, Metrics, or Traces add-on. No separate command required.
Read the docsCentralized log collection, queryable in Grafana with structured filters and labels. Grafana auto-provisions the first time you attach any observability add-on, and stays free.
Useful enough for development, staging, and small production apps. Metered by monthly ingest volume.
Default production grade. HA ingest, 30-45 day retention, generous monthly volume.
TB-scale ingest, 60-180 day retention, ingest replication factor 3, wider query topology.
watasu addons:create logs --app my-app
Read the docs
Scrape Prometheus-style metrics from your processes, query in Grafana, build dashboards and alerts. Platform metrics (pod CPU/memory, restarts, OOMKills, add-on health) are auto-published alongside.
Real custom metrics for development and small production apps. Metered by active series + sustained samples per second.
Production tier — HA ingest/query/storage, 30-45 day retention, room for meaningful instrumentation.
High-cardinality services, long incident-analysis windows, RF=3 ingest replication.
watasu addons:create metrics --app my-app
Read the docs
OpenTelemetry/OTLP trace collection with TraceQL search and Grafana Drilldown. Correlate with logs and metrics in the same Grafana workspace.
Enough trace volume to leave instrumentation on for development and small production. Metered by monthly ingest.
Production tracing — 30-45 day retention, continuous sampling, query fan-out for incident work.
TB-scale ingest, 60-180 day retention, RF=3 ingest replication, separate compactor and query headroom.
watasu addons:create traces --app my-app
Read the docs
You top up a balance, then your usage draws against it. That predictability on our side is a big part of why our prices are this low — and on your side, you can't get hit with a runaway bill.
You get warnings well before zero. If it does run out, your apps pause until you top up — they aren't deleted. No surprise bill shows up in your inbox.
We're a European company billing in our home currency. The price you see on this page is the price your card gets charged — no FX layer in between.
No. Bandwidth is included with your plan, sized to match — normal app traffic at any tier isn't metered, shaped, or invoiced per GB. The only case where we'd reach out is genuinely abusive use (crypto mining, being part of a DDoS), and even then it's a conversation, not a surprise charge.
Yes. Standard 1× at €15/mo runs real production traffic at startup scale. Multi-DC kicks in once you scale to two replicas. For a mostly quiet process, Dynamic 1× stays reachable and bills measured CPU and memory instead.
€5. Usage draws down your balance as it runs. A Hobby app costs €8 for a full month; €5 covers part of the month.
Top up your balance. Upgrade per add-on as you grow. Walk away with your container image any time.