Pricing — hosted, self-hosted or sovereign

Pricing that follows the work, not your headcount.

A platform fee for each connected workspace, plus what the agents actually do in it. The private beta is free, and the teams in it become founding customers — with founding-customer pricing when we launch.

Get early access → Talk to us about pricing

A workspace is one connected platform — your repository, warehouse and orchestrator. Usage is the work the agents do in it: the runs behind a pull request, an investigation, a scheduled check.

Where it runs

Hosted, self-hosted, or entirely inside your borders.

The pricing model is the same in all three. What changes is who runs the infrastructure, and where your data may sit.

Hosted

We run it for you

Managed service, upgrades included.

We run the app, the database and the sandboxes — on Google Cloud in the United States today.
Need another region or cloud? We’ll run your deployment where you need it.
Upgrades, backups and infrastructure are ours to look after.
Agents across the lifecycle: modeling, orchestration and on-call triage.
Every change arrives as a pull request your team approves.
Your own model provider key, and support from us during the beta.

How we run it

Self-hosted

You run it

Your Kubernetes, your database, your network.

A Helm chart on your own Kubernetes, with your own Postgres.
Nothing leaves your network except the calls you configure.
The same agents across the lifecycle: modeling, orchestration and on-call triage.
Every change arrives as a pull request your team approves.
Your own model provider key, and support from us during the beta.

Install guide

Sovereign & regulated

Sovereign deployment

Your region, your network, your model.

For banks, public sector and regulated teams: the same self-hosted product, deployed where your rules require it.
Nothing leaves your network but the calls you configure — no telemetry, usage reporting, licence check or call-back to NerveStax.
Point it at any OpenAI-compatible endpoint — a model you host, or a regional model service — so prompts stay inside your boundary. The keys are yours; we never proxy those calls.
Secrets encrypted with a key that exists only in your deployment.
You choose the region and the operator — whether you run it, or we run a dedicated deployment in the cloud and region you name.

Tell us your constraints

During the private beta, all three are free.

What you pay for

What you’re charged for, and what you’re not.

Charged

Connected workspaces. A platform fee for each platform you connect.
Agent usage. The runs and the work the agents do in those workspaces.

Not charged

Model tokens. You bring your own provider key, and they bill you for the calls.
Reviewers and approvers. We don’t plan to charge per person who reviews a pull request.
Your warehouse, orchestrator and cloud. Those vendors bill you, exactly as they do today.

How we’ll agree pricing

Three steps, in this order.

Talk about your platform

What you run, where the time goes, and which parts of it you’d want an agent near.

Beta, free

You join the private beta at no cost, and become a founding customer while you’re in it.

A plan that matches

Before anything is charged, we agree a plan sized to your workspaces and usage, at founding-customer pricing.

Questions

The things teams ask us first.

Q01Is the private beta really free?+

Yes. No charge to join the private beta, and no card. You bring your own model provider key, and that provider bills you for the model calls.

Q02What happens to the price at launch?+

Beta teams become founding customers and get founding-customer pricing. We will agree your plan with you before anything is charged — nothing starts billing on its own.

Q03Do I pay for model usage through you?+

No. You add your own provider key and the calls go straight to them, under your agreement. We do not resell model usage, and there is no NerveStax fallback key.

Q04Does hosted cost the same as self-hosted to start?+

Both are free during the private beta, and the model is the same either way: a platform fee per connected workspace, plus agent usage.

Q05Can we move from hosted to self-hosted later?+

Yes. It is your repository and your code, and self-hosted is the same product deployed on your Kubernetes. Tell us and we will help you move.

Q06Do you only run in the US?+

The shared hosted service runs on Google Cloud in the United States today. NerveStax deploys as a Helm chart on Kubernetes, so we can stand up a dedicated deployment in another region or cloud — or you can run it yourself. Ask us.

Q07Can we keep everything in our own country and network?+

Self-hosted, yes. It runs on your Kubernetes, your database and your network, and nothing leaves it but the calls you configure. Point it at any OpenAI-compatible endpoint, including a model you host. Secrets are encrypted with a key that exists only in your deployment. Tell us your constraints.

Q08Do you charge per seat?+

We do not plan to charge per person who reviews or approves a pull request. Pricing follows connected workspaces and agent usage instead.