Home / Products

Calimero Confidential Cloud

Compute on data nobody can see.

Sixteen ready-made products for computing on, storing, moving and proving things about sensitive data, on one confidential engine: attested nodes, replicated three times and kept online. Nobody, including us, can read data inside a sealed node.

  • $0to start: a free allowance every month
  • 0servers for you to provision or patch
  • 3attested replicas behind every app instance
  • 16products on one confidential engine

The products

One confidential engineattested nodes · one ordered log per app · sealed channel · signed receipts · fixed governance
Runs on Calimero Cloud: shared clusters or a dedicated clusteror on a cluster of your own

Products compose: a Pipeline can tokenize through the Vault, forward through the Gateway and log to the Audit Log, all on the same engine.

02 · One promise

Proof you can check, built into every product.

Sensitive data goes in, and only what a stated policy allows comes out. You, the people in your data and your regulators can check it from hardware attestation and signed receipts, instead of taking our word for it.

  • Attested hardware

    Data lives only on nodes whose image your client verifies before sending anything.

  • Sealed end to end

    Traffic is bound to the attested key and storage is sealed. Not even the operator can read it.

  • Code you can identify

    Every product is a signed package. You can see exactly which code holds your data, and where.

  • Signed receipts

    Every answer is signed by an attested key and can be checked offline, later, by anyone.

  • Customer consent

    New code, server images and outbound addresses reach data only after approval with a passkey.

  • One ordered log per app

    Replicated across three nodes in one order: no races, no double-spent budgets or approvals.

04 · Starter apps

Start from a working app.

Each starter is a signed app you deploy to Calimero Cloud and adapt: the fastest way to see a use case running on attested hardware.

  • Confidential Agents

    Finance operations agent

    An agent prepares payments; a person confirms with a passkey before money moves.

    Get this starter
  • Vault

    Card tokenization

    Card numbers never touch your servers, shrinking PCI scope.

    Get this starter
  • Gateway + Vault

    LLM privacy proxy

    Prompts reach your AI provider with PII replaced; answers come back restored.

    Get this starter
  • Audit Log

    AI decision records

    Every model decision logged with a verifiable receipt, for the EU AI Act.

    Get this starter

05 · What you get

Everything it takes to keep it online.

We run the nodes and their upgrades. You keep the apps, the data and the proof.

  • 01

    Always on

    Each app instance is replicated by its own Raft log across three attested nodes, so it keeps serving when a node is lost.

  • 02

    Staging and production

    Deploy a version to staging, promote it to production, and roll back when you need to.

  • 03

    Your packages, signed

    Publish signed packages to Calimero's registry, public or private. The signer is pinned at first publish.

  • 04

    Shared or dedicated

    Start on shared clusters and pay per request, or ask for a dedicated cluster at a flat monthly fee.

06 · Verify

Don't trust us. Verify us.

Production images are read-only and have no SSH. A node is admitted only after it proves which image it booted, and your clients check that proof again before they send anything.

  • Keys only after the quote matches the signed release.
  • TLS keys made inside the node, certified only after its quote checks out.
  • Clients verify the node's quote before they open a sealed channel.
The five checksShow less
  1. Signed release

    Measurements published

  2. Node boots

    Intel TDX, read-only image

  3. Node proves

    A quote over its release, bound to a fresh nonce

  4. Key service

    Checks it against release policy

  5. Keys released

    Data unlocked inside the node

A node that fails any step never gets keys. Anyone can check step 3 with the open-source verifier, and the JavaScript, Python and Rust clients and meroctl check every node they talk to.

07 · Pricing

Every product, the same meters.

Two ways to pay, on the same attested nodes. No per-product price lists: whichever products you use, you pay for calls, compute and storage, and every app instance runs three replicas either way.

Request-based

$0 to start

The default, on Calimero's shared clusters. Pay for calls, compute and storage beyond a monthly free allowance. No monthly minimum.

  • 1M mutations and 5M views free each month
  • 180,000 vCPU-seconds and 360,000 GiB-seconds free
  • 1 GB of app state and 5 GB of blobs free
Start free

Dedicated cluster

$1,200/month

Instance-based: a cluster for your apps only, provisioned and run by Calimero. No per-request, CPU or memory fees.

  • 3 nodes: 4 vCPU, 16 GB, 100 GB disk each
  • Extra nodes $0.55 per node-hour, billed per second
  • Usage on it is not metered
Ask for a dedicated cluster
Request-based prices, per month
MeterPriceFree each month
Mutations$1.00 per million1M
Views$0.20 per million5M
App CPU$0.000040 per vCPU-second180,000 vCPU-seconds
App memory$0.0000040 per GiB-second360,000 GiB-seconds
Outbound HTTP requests$1.00 per million100,000
App state$0.50 per GB-month1 GB
Blobs$0.10 per GB-month5 GB

The free allowance is per developer account. Prices are in US dollars.

What each meter counts, and an exampleShow less

Mutations are calls that change an app's state, plus the runs the node makes for your app, such as event handlers and timers; views are read-only calls. App CPU and app memory are the time your app's WebAssembly runs, and the memory it holds while it runs. Outbound HTTP requests are the calls your app makes to hosts you approved. App state and blobs are billed on their average size over the month. Each node meters the apps it runs, and a call is counted once, on the node that ran it. State is billed once, not once per replica.

An example. An app with 3M mutations, 20M views, 400,000 vCPU-seconds, 800,000 GiB-seconds, 2 GB of state and 10 GB of blobs in a month costs $2.00 + $3.00 + $8.80 + $1.76 + $0.50 + $0.50 = $16.56 on shared clusters.

Which model? A dedicated cluster pays off when your apps keep its nodes busy most of the day. The console's usage page prices your month both ways and tells you which is cheaper.

08 · Questions

Before you sign in.

Can Calimero read my data?

No. Nobody, including us, can read data inside a sealed node, and you can verify which image a node runs before it gets any keys. We see what it takes to run and bill the fleet: health, and per-app counts of calls, CPU and memory time, outbound requests and storage size.

Do I pay separately for each product?

No. Every product runs on the same meters: calls, compute, memory, outbound requests and storage, with one free allowance every month.

Can I combine products?

Yes. Products compose on one engine: a Pipeline can tokenize through the Vault, forward through the Gateway and log to the Audit Log.

What is free?

Each month, every developer account gets 1M mutations, 5M views, 180,000 vCPU-seconds, 360,000 GiB-seconds, 100,000 outbound requests, 1 GB of app state and 5 GB of blobs on shared clusters. You pay only for usage beyond that.

Do three replicas cost three times as much?

No. A call is metered once, on the node that ran it, and app state is billed once for all its replicas.

Shared clusters run on spot capacity. Is that safe?

Yes. Each app instance is replicated across three attested nodes, so it keeps serving when one is reclaimed, and a replacement node gets keys only after it proves its image.

When does a dedicated cluster make sense?

When your apps keep a cluster busy most of the day, or a customer wants nodes no one else uses. It costs $1,200 a month for three nodes, plus $0.55 per hour for each extra node, with no per-request fees. The console's usage page compares both models on your real usage.

Where do nodes run?

In Intel TDX confidential VMs with hardware attestation, provisioned and operated by Calimero, on shared clusters or on a dedicated cluster of your own.

Can I run nodes myself?

Yes, for development: merod local runs a three-node cluster on your machine. The node is open source, and core's Operate docs describe a production deployment on your own TDX hosts. Talk to us if a customer needs that.

What languages can I build in?

App logic is written in Rust or JavaScript and compiled to WebAssembly. Client libraries for JavaScript, Python and Rust connect your frontend or service to it.

See your use case running on attested hardware.

Book a technical walkthrough and we'll map your data flow to the right products, or start free on Calimero Cloud.