Starter
Developers
Everything you need to prototype against frontier models.
- Serverless inference
- Shared model catalog
- Community support
- Usage analytics
Pricing
Start on a laptop. Scale to a sovereign estate.
The platform is priced by what you consume and what you need governed. The delivered outcome is scoped per use case — because that is honest work with a target attached, not a seat licence.
Professional
Growing Teams
Dedicated capacity and the controls a real product needs.
- Dedicated endpoints
- Fine-tuning
- Role-based access
- Priority support
- Evaluation suites
Enterprise
Large Organisations
Private networking, governance and procurement-grade terms.
- Private cloud & VPC peering
- SSO, audit and policy
- Committed GPU capacity
- Availability terms agreed per estate
- Named architect
Sovereign AI
Governments
National-scale compute with full data and operational sovereignty.
- Air-gapped deployment
- In-country residency
- Cleared personnel where required
- Bespoke model estate
Delivered outcomes
How a use case is scoped
The tiers above price the platform. A delivered use case is scoped separately, and deliberately so — you are buying work with a number attached rather than access to a console.
We agree the target metric with you up front, and it becomes the definition of done. We will tell you plainly whether the use case is one that ships today as a reference implementation to be tuned on your data, or one we build to order — and we scope the difference honestly rather than quietly.
A shipping use case typically goes from kickoff to a governed, monitored endpoint in weeks. What follows go-live — drift monitoring, evaluators, retuning — is part of the engagement rather than a support contract bolted on afterwards, because an outcome is a durable state and not a launch event.
Questions
What people ask before procurement does
Do we pay for GPUs we are not using?
No. Compute is consumed rather than reserved unless you specifically want committed capacity, which enterprises under a fixed budget frequently do. CPU serving covers a large share of deterministic use cases and costs a fraction of GPU capacity — we will tell you when a use case does not need a GPU.
What happens to our models if we leave?
You keep them. The model, the weights, the governed data it learned from and the decisions it has made are inside your estate throughout — leaving is a matter of us stopping work, not of you extracting anything. Exit terms are agreed up front rather than negotiated under pressure later.
Is the platform priced separately from the delivery work?
Yes. The platform is a subscription with consumption on top; a delivered use case is scoped against its outcome. Keeping them separate is what stops the delivery work from being priced as though it were software licensing.
Do you hold the certifications our vendor review will ask for?
We describe our controls plainly and do not claim certifications we do not hold. Ask us for the current assurance position and you will get the real answer, including where it is incomplete — see the Trust Centre for how we state it.
Talk to us
Get a scoped number, not a price list
Tell us the outcome you want moved and the constraints you are under, and we will come back with what it would take — including if the answer is that you do not need us for it.