Get started
Overview
Nistro AI delivers an enterprise-grade, globally available, open-source AI stack, fully managed for you. Choose what to run and where to run it—including on your own infrastructure.
Our optimized AI stack delivers accurate, high-quality results at a lower cost. With transparent, compute-based pricing instead of per-token charges, you get greater predictability, control, and cost efficiency as you scale.
Why Nistro AI#
- Cost-effective pricing. Nistro AI finds the best-priced GPUs and machines across clouds and regions for every deployment. You pay for the infrastructure you run, not per token. See pricing.
- Your cloud or ours. Run on Nistro AI, or inside your own AWS, Google Cloud, Azure or Nebius account with the Enterprise plan, next to the rest of your workloads and on your existing cloud commitments. See Cloud accounts.
- Open models, familiar API. Deploy any Hugging Face model behind an OpenAI-compatible endpoint. Your existing OpenAI code works by changing the base URL and key.
Products#
| Product | What it does |
|---|---|
| Nistro AI Inference | Run open-weight AI models through an OpenAI-compatible endpoint, optimized for performance, accuracy, and cost. Use Nistro AI-provided GPUs or bring your own GPUs, while Nistro AI handles deployment, scaling, and ongoing operations. |
| Nistro AI Postgres | Fully managed PostgreSQL with automated backups, high availability, scaling, and extensions. Deploy on shared or dedicated infrastructure, or in your own cloud account. |
| Nistro AI Compute | Access NVIDIA and AMD GPUs across multiple clouds and regions. Reserve GPU capacity for two or three years to optimize for performance, availability, and cost. |
Plans#
Inference and Postgres each come in three plans. You pick the plan when you create a deployment, and compare plans on the pricing page.
| On-Demand | Scale | Enterprise | |
|---|---|---|---|
| Best for | Getting started fast with cost-effective pricing | Production workloads on dedicated machines | Running in your own cloud account |
| Where it runs | Nistro AI, in the US, EU or Asia | Dedicated machines in the cloud, region and instance type you choose | Your own AWS, Google Cloud, Azure or Nebius account |
| Placement | Chosen for you by Nistro AI | You choose | You choose |
- On-Demand is the quickest way to start. Choose a model or a database size and a location, and Nistro AI places it on the best-priced capacity.
- Scale adds dedicated capacity and full control over placement. Inference Scale adds dedicated and multi-GPU machines with multiple replicas. Postgres Scale adds high availability, read replicas, backups, point-in-time recovery, snapshots, extensions and parameter tuning.
- Enterprise runs your deployments inside your own cloud account, and Nistro AI still provisions and operates them. Connect the account under Cloud Accounts, then choose Enterprise when you create a deployment. See Cloud accounts.
Get started#
- Sign up and add a payment method. See Getting started.
- Deploy a model or create a Postgres database from the console.
- Call your model with the OpenAI SDK or connect with
psql.
Key concepts#
- Deployment: An instance of a service or workload managed by Nistro AI. Each deployment has its own configuration, status, and management details in the console.
- Team: A workspace for organizing deployments and managing access. Every account starts with a default team, and you can join other teams when invited. Each deployment belongs to a team. See Teams and roles.
- Location: Where your deployment runs. For On-Demand plan, you choose a geographic location (US, EU, Asia, or Any). For other tiers, you can select a specific cloud region.
- Cloud account: Your own AWS, Google Cloud, Azure, or Nebius account connected to Nistro AI to run services within your infrastructure. See Cloud accounts.
- Billing: Charges are based on metered infrastructure usage rather than per-token pricing, giving you greater cost transparency and control. See Usage and billing.
User experience#
The console runs at nistro.ai/console where you create and manage deployments. To work from your terminal or CI, use the Nistro AI CLI.
Next steps#
- Getting started: create an account and deploy your first model or database.
- Nistro AI Inference: deploy a model and call it with the OpenAI SDK.
- Nistro AI Postgres: create, connect to and run databases.
- Nistro AI Compute: GPU capacity reserved for two or three years.
- Platform features: teams, cloud accounts, alerts, usage, billing and support.
- Nistro AI CLI: create and manage deployments from your terminal.