Pay for what runs.
Nothing else.
Every price below is a monthly rate — but you're billed by the minute, only while a service is actually running. No seats. No egress bills. No minimums.
Metered by the minute
Charges accrue every minute a service runs and stop the moment it doesn't. Deploy at noon, delete at 3pm, pay for three hours.
Priced on memory
One simple unit: $3 per 512 MB of hard-reserved RAM per month. CPU comes with it — no separate vCPU meter to decode.
No surprise line items
Bandwidth, TLS certificates, custom domains, team members, builds and deploys are all included. You're billed for compute and storage — that's it.
Every service type, in full.
Pick a size when you create a service, change it whenever you like. Prices are USD per month at full utilisation — run something for half a month and you pay half.
Web services, workers & private services
Your apps and APIs — deployed from Git, a Dockerfile, or a direct upload. Shared burstable CPU scales with the tier, so short spikes ride out without a bigger plan.
| Plan | Memory | CPU | Price |
|---|---|---|---|
| Heli S1 | 512 MB | Shared, burstable | $3/mo |
| Heli S2 | 1 GB | Shared, burstable | $6/mo |
| Heli S3 | 2 GB | Shared, burstable | $12/mo |
| Heli S4 | 4 GB | Shared, burstable | $24/mo |
| Heli S5 | 8 GB | Shared, burstable | $48/mo |
Managed databases
Postgres, MySQL, MariaDB, MongoDB and Qdrant — provisioned, credentialed and connectable in seconds. CPU is reserved (never oversubscribed), so your database is never starved by a noisy neighbour. Postgres ships with pgvector and PostGIS built in.
| Plan | Memory | CPU | Storage included | From |
|---|---|---|---|---|
| Heli DB-S1 | 512 MB | 0.5 vCPU reserved | 2 GB | $3.50/mo |
| Heli DB-S2 | 1 GB | 1 vCPU reserved | 4 GB | $7/mo |
| Heli DB-S3 | 2 GB | 2 vCPU reserved | 8 GB | $14/mo |
| Heli DB-S4 | 4 GB | 4 vCPU reserved | 16 GB | $28/mo |
Redis & key-value
In-memory cache, queue or session store. Priced on memory alone — no storage line, with reserved CPU so latency stays flat.
| Plan | Memory | CPU | Price |
|---|---|---|---|
| Heli KV-N | 256 MB | 0.25 vCPU reserved | $2/mo |
| Heli KV-1 | 512 MB | 0.5 vCPU reserved | $3/mo |
| Heli KV-2 | 1 GB | 1 vCPU reserved | $6/mo |
| Heli KV-3 | 2 GB | 2 vCPU reserved | $12/mo |
ClickHouse analytics
Column-store analytics for events, metrics and logs — with roomier default storage, because analytical workloads land bigger.
| Plan | Memory | CPU | Storage included | From |
|---|---|---|---|---|
| Heli CH-S1 | 1 GB | 1 vCPU reserved | 8 GB | $8/mo |
| Heli CH-S2 | 2 GB | 2 vCPU reserved | 16 GB | $16/mo |
| Heli CH-S3 | 4 GB | 4 vCPU reserved | 32 GB | $32/mo |
Static sites, cron jobs & object storage
The small pieces that surround an app — priced like small pieces, not like servers.
| Service | What you get | Price |
|---|---|---|
| Static site | Built from your repo and served with automatic HTTPS + SPA routing | $1/site/mo |
| Cron job | Scheduled runs, including 60 run-minutes each month | $1/job/mo |
| Object storage | S3-compatible bucket, provisioned 1–64 GB as a hard quota | $0.08/GB/mo |
Optional add-ons
Only ever charged when you switch them on.
| Add-on | What it does | Price |
|---|---|---|
| Block storage | Extra database volume, adjustable 2–64 GB at any time | $0.25/GB/mo |
| Daily backups | Automated daily database backups | +20% of storage |
| Dedicated CPU | Guaranteed cores instead of shared burstable | $3/0.5 vCPU/mo |
Real stacks, real totals.
No calculator required — here's what a typical setup comes to at full-month utilisation.
App + database
API, database & cache
Two services + data layer
The platform is free. You pay for compute.
Most clouds meter the conveniences and bill you for the privilege of using them. We don't. Everything below ships with every account, at every size, at no extra cost.
Domains & networking
- Custom domains — add as many hostnames to a service as you need, at no charge
- Automatic HTTPS on every hostname, issued at first handshake and renewed for you
- An instant public HTTPS subdomain the moment a service first deploys
- Private service networking — internal-only services never touch the public internet
- Static outbound IPs to allowlist in third-party APIs and databases
- No bandwidth or egress charges, ever
Builds & deploys
- Push-to-deploy from GitHub — merge and it ships
- Zero-config builds for Next.js, Nuxt, SvelteKit, Vite, Go, Rust, Python, Ruby, PHP, Java and Node
- Bring a Dockerfile, a public or private image, or drag-and-drop a folder — no git required
- Zero-downtime blue/green releases with a health-checked candidate before traffic moves
- One-click rollback to a previous release, no rebuild
- Release commands & deploy hooks — run migrations, verify a build before it takes traffic
- Unlimited builds and deploys. We don't meter build minutes
Reliability
- Health checks with auto-restart — a wedged container is replaced before you notice
- Crash-loop protection that stops the bleeding and emails you the likely cause
- Scale-to-zero services that cost nothing while idle and cold-start on request
- Hardened containers — dropped capabilities, no-new-privileges, fork-bomb caps
- Blocked cloud-metadata egress & SSRF guards on every outbound path we control
- Automatic log rotation and disk reclamation so a chatty app can't fill the host
Logs, metrics & access
- Runtime logs for apps and databases, with real full-text search
- Live-streaming deploy logs, and your app's own ANSI colors preserved
- CPU, memory and network metrics with p50/p95/p99 and 1h–7d ranges
- Engine-native database metrics — connections, cache hit ratio, ops/sec and more
- An interactive shell straight into a running container, in the browser
- Log drains to Papertrail, Better Stack, Datadog, syslog or Vector
Environments & data
- Dev, staging and production environments — isolated, per project
- Duplicate an entire environment in one action, references remapped for you
- Env groups & secret files shared across services, mounted read-only
- Service references that wire
DATABASE_URLautomatically and re-resolve on every deploy - Secrets encrypted at rest with AES-256-GCM, masked by default, audited on reveal
- A data explorer for every engine — tables, collections, keys, vectors and buckets
- On-demand backups and one-click restore that runs inside the container
Team, tooling & migration
- Teammates at no extra cost — we don't charge per seat, ever
- Per-project access control, so contractors see only what they should
- API keys and a hosted MCP endpoint — point Claude or Cursor at your infrastructure
- The heli CLI for deploys, logs, scaling and status from your terminal
- One-click migration from Railway, Render, Vercel or a docker-compose file — with your data
- Templates and blueprints — n8n, Evolution API and full database stacks in a click
- 7-day trash with true restore — a deleted service comes back with its data
The fine print, in plain English.
What does “billed by the minute” actually mean?
Is there a free tier?
Do you charge for bandwidth or egress?
Can I change plans later?
What happens if a service is idle?
How do payments and invoices work?
Start with one service. Scale when it earns it.
Connect a repo and go live in minutes — on a platform that bills you by the minute and includes everything else.