Calculator

Estimate your private AI costs

Get an indicative monthly figure from our published plan prices, then compare it with running the same capability yourself. Both tools run entirely in your browser.

Start estimate Build vs. buy

Cost estimator

What would your deployment cost?

Choose a deployment, model size and expected load. The estimate uses our current plan prices — it is not a binding quote.

Estimate only

Figures are indicative. A binding quote follows an architecture review, where we validate model fit, context length and capacity against your workload.

Deployment type

Managed API (shared)Shared inference behind an OpenAI-compatible API.Dedicated nodeSingle-tenant node, managed for you.High availabilityTwo-node architecture with load balancing.

Model size class

SmallUp to ~14B parameters, embeddings, reranking, speechMedium24–32B class (dense or mixture-of-experts)Large70B class, quantizedVery large~120B mixture-of-experts

Expected usage

Peak concurrency

Storage needs

Support level

High availabilityTwo nodes with load balancing and a failover design.Add a Mac CI nodeSeparate Apple Silicon node for iOS / macOS builds.

Your estimate

Estimate — not a binding quote

Estimated monthly total

from €699 / month

≈ €583 / month with annual billing. Annual billing: 2 months free, invoiced yearly in advance.

Managed AI Nodefrom €699

Deployment: Dedicated node
AI nodes: 1

Recommended architecture

1 × dedicated 96 GB node · private API endpoint · managed deployment and monitoring

A single dedicated 96 GB node, managed for you.

Running your own model stack? The Dedicated Node starts from €449 / month.

Get a binding quote

Indicative only, based on our published plan prices. Node counts are a starting point — capacity is validated against your workload before we quote. All prices exclude VAT.

Build vs. buy

Self-hosting or managed infrastructure?

Hardware is only part of the cost of running models yourself. Enter your own figures to compare an in-house setup with a Managed AI Node.

Monthly comparison

Self-hosting (amortised)€2,341

Managed with Lirux€699

Difference per month

€1,642

Lower with managed infrastructure, under these assumptions.

≈ €19,705 per year.

Self-hosting cost breakdown per month
Self-hosting item	€ / month
Hardware depreciation€8,000 over 36 months	€222
Replacement reserve10% of hardware per year	€67
DevOps time10 h × €90	€900
AI/ML engineering time8 h × €110	€880
Electricity & coolingper month	€60
Maintenance & spare partsper month	€40
Monitoring & toolingper month	€50
Setup engineering (amortised)€4,400 one-time over 36 months	€122
Self-hosting total	€2,341
Managed total	€699

One-time setup engineering: €4,400, valued at the AI/ML engineering hourly cost and spread over the depreciation period. Application integration work is similar in both cases and is not included.

Results depend entirely on the assumptions above and are not a statement of fact about any particular organisation.

How pricing works

Predictable by design.

Fixed monthly plans

Shared API plans start at €49 and dedicated nodes at €449 per month. You pay for capacity, not per token.

Sized to your workload

Model size, request volume and concurrency decide whether shared capacity is enough or a dedicated node — or two — is the better fit.

Add-ons quoted, not guessed

Extra storage, nodes beyond a two-node setup and Enterprise support are priced on request after an engineer reviews your architecture.

Monthly or annual

Annual billing: 2 months free, invoiced yearly in advance. All prices exclude VAT.

Production SLA options are available for eligible managed deployments.

Get a binding quote Compare all plans

Turn the estimate into an architecture.

Share your model, volume and constraints. An engineer will come back with a concrete deployment proposal and a binding price.

Design my infrastructure View pricing