Pricing · 6 server sizes · Billed monthly

One server. Everything on it.

Each plan is a server that Aria sets up and runs for you, and the price covers both. Prices start from the lowest location available now, and each location’s price is shown on the card. Run as many apps and databases as it can hold. There are no charges per app, per seat or per request.

From£8a month, ex VAT

Draft prices. They may change before launch. VAT is added at checkout where it applies.

  • 6server sizes
  • up to16vCPU
  • up to32GBmemory
  • up to400GBstorage

The plans

Six sizes. One price each.

Every plan is one server Aria sets up and runs. The figure on a card is the lowest location available now; open “Price by location” to see each one.

01

Starter

One app with a database.

Good forA first app: a site or an API with one database.

  • Landing page + API
  • Small SaaS
  • Side project

from£8a month, ex VATDraft price

Compute
2 vCPU
Memory
4 GB
Storage
40 GB
02Supabase ready

Developer

A few services, Supabase, room for builds.

Good forA real product, or self-hosted Supabase with room for builds.

  • Next.js + Supabase
  • API + worker
  • Staging server

from£15a month, ex VATDraft price

Compute
4 vCPU
Memory
8 GB
Storage
75 GB
03Supabase ready

Pro

Bigger stacks and heavier traffic.

Good forBigger stacks: several services, queues and heavier traffic.

  • AI app + vector DB
  • Web + API + workers
  • Busy SaaS

from£22a month, ex VATDraft price

Compute
6 vCPU
Memory
12 GB
Storage
100 GB
04Supabase ready

Business

Several apps or a busy one, with room to grow.

Good forSeveral apps for one company, or one busy app with room to grow.

  • Agency client apps
  • 3–5 production apps
  • Internal tools

from£30a month, ex VATDraft price

Compute
8 vCPU
Memory
16 GB
Storage
200 GB
05Supabase ready

Scale

Many services, large databases, busy traffic.

Good forLarge databases and many services under steady load.

  • Marketplace
  • Analytics workloads
  • Many microservices

from£40a month, ex VATDraft price

Compute
8 vCPU
Memory
24 GB
Storage
200 GB
06Supabase ready

Scale Plus

The biggest apps before a dedicated server.

Good forThe biggest apps that still fit on one cloud server.

  • High-traffic product
  • Large Postgres
  • Monorepo of services

from£60a month, ex VATDraft price

Compute
16 vCPU
Memory
32 GB
Storage
400 GB

Dedicated servers · Planned

When one cloud server isn’t enough.

A whole machine, not a slice of one: every core, every gigabyte and both disks are yours alone, with the same Aria on top. Dedicated servers are coming soon; these are the sizes and the prices we plan to charge.

Planned

Metal 1

Your first whole machine.

£100a month, ex VAT

Processor
Intel Xeon-E · 6 cores / 12 threads
Threads
12
Memory
32 GB
Storage
2 × 512 GB NVMe
Planned

Metal 2

Twice the memory for big databases.

£170a month, ex VAT

Processor
Intel Xeon-E · 6 cores / 12 threads
Threads
12
Memory
64 GB
Storage
2 × 512 GB NVMe
Planned

Metal 3

AMD EPYC with fast DDR5 memory.

£260a month, ex VAT

Processor
AMD EPYC 4344P · 8 cores / 16 threads
Threads
16
Memory
64 GB
Storage
2 × 960 GB NVMe
Planned

Metal 4

Twelve cores for the heaviest workloads.

£350a month, ex VAT

Processor
AMD EPYC 4464P · 12 cores / 24 threads
Threads
24
Memory
64 GB
Storage
2 × 960 GB NVMe

Need room today? Scale Plus is the largest cloud server, ready now.Start on Scale Plus

GPU servers · On request

For apps that need an NVIDIA GPU.

AI models, image and video generation, transcoding and 3D rendering. Mark the service withgpu: true in aria.yaml, or give the GPU memory it needs. GPU servers are on request for now: ask from the app and the Aria team emails you a proposal with the server and its monthly price. These are examples of what we can provide. Apps that only call a hosted AI API don’t need one.

GPU 16

An NVIDIA GPU with 16 GB, for small AI models, Whisper and video work.

On request

GPU
NVIDIA V100, 16 GB
Processor
8 vCPU
Memory
45 GB
Storage
300 GB NVMe

GPU 24

An NVIDIA L4 with 24 GB, for 7–13B models, image generation and transcoding.

On request

GPU
NVIDIA L4, 24 GB
Processor
22 vCPU
Memory
90 GB
Storage
400 GB NVMe

GPU 48

An NVIDIA L40S with 48 GB, for big models, video generation and 3D rendering.

On request

GPU
NVIDIA L40S, 48 GB
Processor
15 vCPU
Memory
90 GB
Storage
400 GB NVMe

Interactive sizer

What are you deploying?

Pick the stack closest to yours. The memory figures are estimates; the matching plan lights up above.

Select a workload to highlight the matching server tier

In every plan

What one server holds.

  1. 01 / 06

    Apps & databases

    As many apps, APIs, workers and databases as the server can hold

  2. 02 / 06

    Postgres & Redis

    PostgreSQL and Redis, with backups on the server

  3. 03 / 06

    Logs & rollback

    Build and runtime logs, and one-click rollback

  4. 04 / 06

    Agents over MCP

    Claude Code and other agents over MCP, under the same safety rules

  5. 05 / 06

    AI Operator

    The AI Operator, once it’s configured, with a monthly usage limit

  6. 06 / 06

    Mac app & dashboard

    Native macOS 26 Liquid Glass app and full browser dashboard

How the price works

No charges per seat, app, database or request

One server. One price.

An Aria plan is one monthly price for one server. Run as many apps, workers and databases on it as it can hold, and invite your agents too: nothing is metered per seat, per app or per request.

Answers & billing details

Questions, answered.

Want a side-by-side table? Compare Aria with Vercel, Supabase Cloud, Heroku & Render →

Why are the prices marked draft?

These are the prices we intend to charge, and they may change before launch.

What happens at checkout?

You name your app, choose where its server runs and pick a plan, here or in the Mac app. You pay on Aria’s own checkout page, where the card details go straight to Stripe, never to Aria. Prices don’t include VAT, which is added at checkout where it applies. Your server is ordered the moment the payment goes through.

Where does my server run?

On a dedicated cloud server Aria rents and runs for you, so the one price covers it, and it isn’t shared with anyone else. Ordering, securing and setting up those servers is built and tested; it opens once Aria’s cloud account is connected. Until then, Aria runs against a local test server on a Mac.

Why does the price depend on where my server runs?

A server costs a little different in each place. The price on a plan card is the lowest one available now, and under it you see the price for each location, with the size of the server you get there. You pay the price you chose, and prices are always ex VAT. If a price changes before you pay, Aria shows you the new one first and charges nothing until you agree.

Can I change plan later?

Yes, in the Aria app or on the web. Aria checks memory, CPU and disk every minute and suggests a bigger plan (and emails you) when they run short. A bigger plan grows your server where it can, with one restart. When it can’t, Aria moves your app to a bigger server instead: before you pay, it shows the price and how many minutes the app is down, and your addresses keep working. Plans can’t shrink to a smaller one.

What about Supabase?

Self-hosted Supabase works today on the local test server. It needs more memory than a small app, which is why Developer and Pro and Business and Scale and Scale Plus are sized for it.

How does Aria pricing compare to Vercel, Supabase Cloud, Heroku and Render?

Vercel charges per team seat plus usage; Supabase Cloud charges a base fee plus per-project and database compute add-ons; Heroku, Render and Railway charge per dyno or service plus database fees. Aria charges one flat monthly price for your dedicated server, with £0 per-seat, £0 per-app, £0 per-database and £0 per-request fees.

Start

Ready to launch your first server?

Pick a plan when you start your first app, here or in the Mac app. You can move to a bigger one later, and Aria emails you before memory runs short.

Request the server size

We don’t have this size ready right now. Tell us where you want it: we buy it for you and email a code within a few hours, in working hours. The code is held for 72 hours, and you pay only when you use it.

We’ve updated our terms

Please read the new terms of service and privacy policy. Accept them to keep using Aria.