Skip to main content
All CollectionsCompute
Understand and choose your Compute instance

Understand and choose your Compute instance

A quick guide to picking, launching, and connecting to the right instance for your needs.

Thanasis Karavasilis avatar
Written by Thanasis Karavasilis
Updated this week

What makes up a Compute instance?

Each Compute instance is a container built for high-performance workloads. When you create one, you’re choosing a mix of hardware and settings that define how it performs.

Here’s what’s included:

  • GPU(s): The core of your instance. Choose from NVIDIA RTX 4090s — up to 8 per machine — ideal for training AI models, running ML workloads, 3D rendering, and other GPU-heavy tasks.

  • vCPUs & RAM: These are automatically matched to your GPU setup, so everything runs smoothly without manual tweaking.

  • Storage: Every instance comes with fast local SSD storage. Persistent or expandable storage isn’t available yet, but it’s coming soon.

  • Region: Pick where your instance runs — (for now, we support France or the UAE). Your choice can affect latency and data residency.

  • Billing type: We currently support on-demand billing (pay-as-you-go). You get high availability without long-term commitments.

💡 Tip: You don’t pay when your instance is off. Terminate it when you're done to stop billing.

How to choose the right instance

The right setup depends on what you’re working on. Here are a few tips to guide your decision:

  • Training AI models? Each RTX 4090 has 24 GB of VRAM. Pick the number of GPUs that matches your model’s size and memory needs.

  • Not sure what to choose? That’s okay. Reach out via the chatbot or join us on Discord — we're here to help.

Managing your instance from the dashboard

Once it’s live, your instance is fully under your control:

  • Start/Stop anytime

  • Monitor usage, uptime, and cost in real time

  • Terminate the instance when you're done — billing stops as soon as it’s off

🐝 Heads up: Instances are billed per second, so you’re never stuck paying for idle time.

How to connect to your instance

As soon as your instance is ready, you’ll get:

  • A public IP address

  • SSH credentials

  • GPU details (so you can run nvidia-smi and check your setup)

You can connect using your terminal with a standard SSH command:

ssh [your-username]@[your-instance-ip]


You can use your terminal or a GUI SSH client — whichever you prefer.


Need a hand?

If you're stuck or just want a second opinion, we're happy to help.

Did this answer your question?