
No sharing, no queues
It is not a share of a card split between customers: it is entirely yours for as long as the rental lasts.
From 0,15 €/hour + VAT · No lock-in · Root access via SSH
Not a slice of a shared GPU: you notice it in the latency, and in it never slowing down. No lock-in: you pay for the hours you use.

It is not a share of a card split between customers: it is entirely yours for as long as the rental lasts.

You give us your public key when you order and log in as root. There is no password to remember, or to leak.
We only offer cards that are verified and available to rent at that moment.
No direct debit. When the hours run out, the server shuts itself down.
Just like our other services: no accounts elsewhere, no cards handed to third parties.
Open source, with no reliance on a third-party API or its usage limits. The price is that of the cheapest GPU in the catalogue with the VRAM each model needs.
The go-to open model for chat and assistants.
VRAM ≥ 16 GB From 0,25 €/hourFast and lightweight, widely used in production.
VRAM ≥ 16 GB From 0,25 €/hourReasoning and code, top-tier performance.
VRAM ≥ 24 GB From 0,60 €/hourMultilingual, strong in Spanish too.
VRAM ≥ 16 GB From 0,25 €/hourText-to-image generation.
VRAM ≥ 12 GB From 0,15 €/hourAudio-to-text transcription, in dozens of languages.
VRAM ≥ 6 GB From 0,15 €/hourNeed another model, or your own? It installs the same way: it is your server.
Priced by the hour. Buy as many hours as you like and it shuts itself down when they run out. No direct debit.
12 vCPU · 16 GB RAM · 342 GB NVMe
4 vCPU · 16 GB RAM · 75 GB NVMe
4 vCPU · 16 GB RAM · 184 GB NVMe
8 vCPU · 8 GB RAM · 68 GB NVMe
4 vCPU · 23 GB RAM · 345 GB NVMe
12 vCPU · 63 GB RAM · 260 GB NVMe
24 vCPU · 16 GB RAM · 416 GB NVMe
8 vCPU · 16 GB RAM · 339 GB NVMe
20 vCPU · 16 GB RAM · 162 GB NVMe
96 vCPU · 63 GB RAM · 653 GB NVMe
64 vCPU · 31 GB RAM · 81 GB NVMe
192 vCPU · 63 GB RAM · 93 GB NVMe
192 vCPU · 126 GB RAM · 3 TB NVMe
16 vCPU · 31 GB RAM · 648 GB NVMe
16 vCPU · 31 GB RAM · 140 GB NVMe
80 vCPU · 31 GB RAM · 294 GB NVMe
80 vCPU · 126 GB RAM · 756 GB NVMe
16 vCPU · 63 GB RAM · 800 GB NVMe
64 vCPU · 31 GB RAM · 561 GB NVMe
96 vCPU · 50 GB RAM · 1.3 TB NVMe
32 vCPU · 101 GB RAM · 26 GB NVMe
24 vCPU · 31 GB RAM · 784 GB NVMe
64 vCPU · 31 GB RAM · 1.5 TB NVMe
80 vCPU · 126 GB RAM · 616 GB NVMe
40 vCPU · 63 GB RAM · 525 GB NVMe
28 vCPU · 47 GB RAM · 1.5 TB NVMe
12 vCPU · 93 GB RAM · 14 GB NVMe
32 vCPU · 125 GB RAM · 1 TB NVMe
48 vCPU · 23 GB RAM · 100 GB NVMe
48 vCPU · 47 GB RAM · 88 GB NVMe
32 vCPU · 63 GB RAM · 144 GB NVMe
24 vCPU · 31 GB RAM · 159 GB NVMe
128 vCPU · 84 GB RAM · 710 GB NVMe
16 vCPU · 47 GB RAM · 2.8 TB NVMe
96 vCPU · 126 GB RAM · 2.8 TB NVMe
No model matches that filter. Clear filters.
PriceOn request
Enterprise · Multi-GPU
From 2 to 8× H100 80 GB with NVLink, plus CPU and RAM to suit your project.
No sign-up forms, no waiting: the server is created automatically as soon as payment is confirmed.

From the live catalogue, with the hourly price in front of you. You give us your public SSH key.
As soon as payment is confirmed, the server is created. You are usually in within a few minutes.
When the hours run out it shuts down and deletes itself. If you want to keep going, extend beforehand and there is no interruption.
If yours is not here, write to us and we will answer, with no salesperson in between.
Write to support
Tell us what you need and we will tell you what server you need, in under a day.
We use our own cookies and those of Google and Meta to measure which ads bring in customers. They are only set if you accept; you can reject them and the site works just the same. More details.