Free
Start here$0
$10 of free usage per user each month
For individuals exploring the platform. Anyone can sign up under our fair use policy.
- Chat and the API on open-weight models
- Playground, projects and sources
- Usage dashboard in the app
Pricing
Start free, then choose how your firm buys
Try PrivateMind free under our fair use policy, then agree a contract that fits how your firm buys, with committed spend or dedicated GPUs. Card payments are coming soon.
How pricing works
- Per-model rates quoted on request
- Batch at half price
- $10 of free usage every month
- Committed spend discounts for enterprise
Pay as you go
Coming soonCredits
Pay by card for the tokens you use
Top up credits by card and pay only for the tokens you use. Until card payments launch, talk to us about usage beyond the free allowance.
- Every model in the catalogue
- Batch requests at half price
- Budgets and rate limits on every key
- Live usage and spend in the app
Enterprise
For firmsCustom
committed spend or dedicated GPUs
For firms rolling PrivateMind out across teams, with contracts built around how you want to buy.
- Discounts for committed spend
- Dedicated GPUs and models as a service
- SSO, compliance API and audit logging
- SLAs and a named account team
Model pricing
Quoted per organisation
Per-model rates on pooled capacity are set per organisation. Batch requests are half price, and committed spend receives a discount. Contact us for current rates.
Compute pricing
Dedicated GPUs, priced for your workload
Bare metal NVIDIA GPUs, from a single card or MIG slice to whole pods, with networking, storage and 24/7 operations included. Prices depend on the GPU, the term and the region, so we quote each one.
| GPU | Memory | Architecture | Price |
|---|---|---|---|
| NVIDIA B300 | 288 GB HBM3e | Blackwell Ultra | Contact sales |
| NVIDIA B200 | 180 GB HBM3e | Blackwell | Contact sales |
| NVIDIA H200 | 141 GB HBM3e | Hopper | Contact sales |
| NVIDIA H100 | 80 GB HBM3 | Hopper | Contact sales |
| NVIDIA RTX PRO 6000 | 96 GB GDDR7 | Blackwell | Contact sales |
Fractional GPUs, reserved terms and multi-pod clusters are available. See the infrastructure or request a demo for a quote.
Enterprise
Contracts built around how your firm wants to buy
Most firms combine more than one of these. We will put together a contract that fits your usage, your procurement process and your plans.
Committed spend
Commit to an amount of usage over the term of your contract and receive a discount on token prices across the catalogue.
Models as a service on dedicated GPUs
We buy, host and run the GPUs for you and serve the models your firm needs on them. At high, steady volumes, or once you train your own models, this usually costs less than paying per token, and you never manage any hardware.
Bare metal GPUs
Reserve B300, B200, H200, H100 or RTX PRO 6000 capacity, from a few GPUs to multiple pods, with managed services included.
AI Solutions
Deployed AI engineers and workshops can be added to any contract, scoped to the work you want to achieve.
Questions
Billing questions
Something not covered here? Request a demo or read about budgets and limits in the docs.
How do credits work?
Credit card payments and prepaid credits are coming soon. When they launch, each request will be priced on its input and output tokens at the rates for that model and taken from your balance. Until then, talk to us about usage beyond the free allowance.
What happens when I reach my limit?
Requests return a clear error when you reach your allowance or a budget cap, so usage never runs on unexpectedly. Individual keys can also have their own budget caps.
How is batch priced?
Requests sent through the Batch API are billed at half the normal token rate, in exchange for results arriving within a completion window rather than straight away.
Can we control spend across a team?
Yes. Budgets and rate limits can be set per API key, per user and per organisation, and are enforced on the server.
How are frontier models priced?
If your firm enables hybrid routing, Claude, OpenAI and Amazon Bedrock models are billed through the same balance. Request a demo for current rates.
Can we pay by invoice?
Enterprise contracts are invoiced. Talk to our team about committed spend or dedicated capacity.
Start free today, or request a demo of private AI for your firm
Create an account to use chat and the API straight away, or request a demo to see dedicated capacity, workspaces and engineers for your team.