Choose the hardware
Review the GPU model, memory and host platform first. Listings explain what was checked, so you can judge whether the computer fits your work.
Closed betaGPU compute for your next project
Give your next idea room to run.
Explore whole GPU computers for the hours your project needs, with clear listing details and SSH access using your own key.
- Choose your computer
- Choose your window
- Connect with your key
Closed beta · by invitation
- One renter at a time
- Choose your hours
- SSH with a key
- Clear listing details
- Invitation-only beta
For your next project
Your work. A computer to run it.
Start with the GPU and time your project needs. A rental will give you access to a computer for your rental window, using your own tools.
Make time for your project
Search for the hours you need. No other renter shares the computer's GPUs during your rental window.
Work with your own tools
Connect to the rental environment over SSH with your key. You run your workload and keep your checkpoints and results somewhere outside the rented computer.
Inference
Run the open model you choose.
Bring your own runtime and model over SSH. Your prompts go to the model you run, on a computer no other renter shares.
RTX 4090
24 GB
of graphics memory
Room for a 27B to 35B open model at 4-bit with a working context, or gpt-oss-20b with its full context.
RTX 5090
32 GB
of graphics memory
The same models with much longer context, or a 27B model at 8-bit.
A private chat
Talk to the open model you choose, over your own SSH connection.
An API for your own tools
Serve an OpenAI-compatible API inside the rental, and reach it on your own computer's localhost.
Batch jobs and embeddings
Run a model over a whole dataset, then copy the results out before your window ends.
The rental journey
How renting will work.
Follow the path from choosing compute to reviewing your rental.
Choose your computer
Review the GPU, memory and host platform in the listing, along with the limits and terms that apply.
Choose your window
Search for your start time and duration, and see each computer's price per hour and when it is free.
Connect with your key
When your window starts and your rental is ready, connect over SSH with your key and run your workload.
Finish and review
Save your results before the window ends. Your receipt records the rental; if something went wrong, you can open a dispute from the same place.
Know your rental
Know what you get.
Read the listing, understand the limits and choose a computer that fits your work.
One renter at a time
No other renter shares the computer's GPUs during your rental window. The host still controls the machine and can use or stop it.
Clear details, honest limits
Hardware details are provided by the host. We review listing information, but do not guarantee the hardware or its performance.
Plan for your data
Hosts can technically access data on their computers. Keep backups elsewhere; do not process highly sensitive data on a Windows host.
Before you begin
A few things to know.
What will I need to connect?
You will need a Nalvo account and an SSH public key, and should be comfortable running your own workload over SSH. See the Compute Market for the rental details.
Who owns the computer?
The computer belongs to a host, who can technically access data on it. Do not process highly sensitive data on a Windows host. Read the data limits and Windows host differences before you start.
Can I bring my own computer to Nalvo?
Yes, there is a separate path for owners. Visit Host on Nalvo to learn about managing your computers and choosing whether to offer a machine for rent.
Express your interest
Be part of what comes next.
Host rentals on your GPU PCs, or rent one for your work. Tell us which: it takes about two minutes, and we read every application.
Questions about Nalvo.ai, the closed beta or the Compute Market? Write to the same address.
The privacy notice explains how we handle applications, enquiries and website analytics.
Please do not email passwords, access tokens, card details or sensitive device information. See the privacy notice for how enquiries and website analytics are handled.