Pricing

Priced for infrastructure, not for tokens.

Your hardware does the work. Nodeau's job is making that hardware usable and manageable, so nothing here is metered per prompt, per token or per request.

One machine is Home. A few machines that are all yours is Home Pro. An organisation, where other people need their own access, is Business.

Home

Make my GPU useful.

Free

No card. No account. Yours to keep.

One machine, one GPU
Local inference on your own hardware
Chat, embeddings, reranking, structured output, tool calling and image understanding
The curated model catalogue, with the card size each model wants
Fit checks before anything starts
Efficiency, Balanced and Performance scheduling
Power limits and budgets for your cards
Local OpenAI-compatible API with its own key
The local dashboard
CLI, diagnostics and support bundles

Free means free. Everything that keeps you safe is in every plan and stays there. Model verification, authentication and the fit checks come with all three.

Install Nodeau

Home Pro

Make all my GPUs work together.

$9.99 / month

For one person with more than one machine. Cancel any time.

Everything in Home
Up to three GPU machines under one control plane
Two GPUs in a machine, working on one model or on two
Batch inference on any machine in the fleet, with several workers at once
Remote dashboard and management from anywhere
Drain a machine for maintenance and pick the work back up
Standard support

This is the tier for one person with a desktop, a server in the cupboard and a laptop that wants to reach both of them.

Get Home Pro

Business

Private AI infrastructure for an organisation.

Priced per organisation, not per request.

Everything in Home Pro
Fleets that mix cards and generations, judged on their own merits
A control plane that explains every placement it makes
Business support, and a direct line to the people building it
A say in what organisation-scale operation becomes

Design partners shape how organisation-scale operation gets built. We would much rather build it with you than at you.

Talk to us

What's in each plan

The paid plans are worth paying for because they do more. The free one is a whole product in its own right.

Nodeau plans compared. Limits are per installation.
Capability	Home	Home Pro	Business
The basics, in every plan
Local inference on your own GPU	Included	Included	Included
Local dashboard	Included	Included	Included
Curated model catalogue	Included	Included	Included
Model download, verification and cache	Included	Included	Included
Fit checks, explained refusals, diagnostics	Included	Included	Included
Embeddings, reranking, structured output, tool calling, vision	Included	Included	Included
Efficiency, Balanced and Performance scheduling	Included	Included	Included
Power limits and budgets for your cards	Included	Included	Included
Accept a fit risk yourself, when you want to	Included	Included	Included
Scale and automation
GPU machines	1	Up to 3	Unlimited
GPUs in one machine	1	Up to 2	Unlimited
Workloads at once	2	Unlimited	Unlimited
Batch inference	Starts with Home Pro	Included	Included
See your machines from anywhere	Included	Included	Included
Remote dashboard and management	Starts with Home Pro	Included	Included
Support	Community	Standard	Business

Everything listed here works today. You are only ever charged for what already runs on your hardware, and the roadmap is where we keep what is still being built. Business is something we shape with design partners, so talk to us and we will build it around what you actually need. Plans are enforced by a signed entitlement your installation checks locally, which means everything keeps running while your machine is offline.

Why there is no per-token price

Your GPU does the work

The work happens on hardware you already paid for. Charging per token would be charging you for your own electricity.

The value is management

Deciding what fits, keeping it running, and knowing what is where. That scales with machines, not with prompts.

Capacity you can predict

Owned hardware has a fixed ceiling and a fixed cost. That is a different shape from per-request billing, and for some workloads a better one.

Owned hardware is not automatically cheaper than cloud inference. It depends entirely on utilisation, and the honest answer for a machine that sits idle is usually no.

Start on the machine you already have.

Home is free, needs no account, and will tell you straight whether your machine can run it.

Install Nodeau Talk about Business