Get started
Introduction
What Nistro AI is, its products and plans, and where to go next.
Nistro AI runs managed PostgreSQL databases and open-weight AI models for you. You choose what to run and where it runs; Nistro AI provisions the infrastructure, keeps it running and bills for the infrastructure by the hour rather than per token.
Products#
| Product | What it does |
|---|---|
| Nistro AI Postgres | Managed PostgreSQL. Size a database yourself on a shared fleet, or run it on dedicated machines with standby replicas, backups, extensions and parameter tuning. |
| Nistro AI Inference | Deploys open-weight models from Hugging Face on GPUs and serves them behind an OpenAI-compatible endpoint with a per-deployment API key. |
Plans#
Each product comes in three plans. You pick the plan when you create a deployment.
| On-Demand | Scale | Enterprise | |
|---|---|---|---|
| Where it runs | Nistro AI's hosted fleet, in US, EU or Asia, or the lowest-priced of them with Any | Dedicated machines in the cloud, region and instance type you choose | Your own cloud account |
| Placement | Managed by Nistro AI | You choose it | You choose it |
| Access | Self-serve | Postgres: self-serve. Inference: on request | On request |
- On-Demand is the quickest way to start. You choose what to run, a database size or a model, and a location; Nistro AI places it on its fleet.
- Scale gives you dedicated capacity and control over placement. Postgres Scale adds standby replicas, daily backups, snapshots, extensions and parameter tuning. Inference Scale adds dedicated GPU machines, multi-GPU instances and a fixed number of replicas.
- Enterprise deploys into your own cloud account. It is set up with our team; select Contact us on the Enterprise plan in the console.
The product pages list what each plan supports in detail.
Key concepts#
- Console: the web app at nistro.ai/console where you create and manage deployments.
- Deployment: one database or model. Each deployment has a status, connection details and a detail page in the console.
- Team: every account has its own team, named after you (for example "Sam's Team"). When another account owner shares a plan with you, their team appears in the team switcher too. Deployments belong to the team they were created in. See Teams and roles.
- Location: where an On-Demand deployment runs: US, EU or Asia. With Any, the default, Nistro AI picks the lowest-priced location.
- Billing: charges follow metered usage of the infrastructure you run. You need a payment method before you can deploy. See Usage and billing.
API#
The console is built on the Nistro AI API. The API reference lists its endpoints and their parameters.
Next steps#
- Getting started: create an account and deploy your first database or model.
- Nistro AI Postgres: create, connect to and operate databases.
- Nistro AI Inference: deploy a model and call it with the OpenAI SDK.
- Usage guide: teams, alerts, usage, billing, support and limits.
Questions? Email [email protected].