From a single GPU to a whole cluster

For teams that run their own models: launch a virtual machine with root access, train without managing machines, or reserve a cluster with our engineers.

Start buildingTalk to an expert

Which option fits your workload

Start with an on-demand GPU, then move when your work changes.

  • On demand

    Start and stop when you like

    Root access. Billed per second, no subscription.

    Best for: ExperimentsShort jobsCustom stacks

    See GPUs and prices

  • Spot

    Same GPUs, lower rate

    About 50% to 60% less. Can be reclaimed at any time.

    Best for: Checkpointed jobsBatch evaluationsReruns

    Compare spot prices

  • Serverless training

    No machine to look after

    Your container, on the GPU you choose.

    Best for: Fine-tuningBatch jobsRepeatable runs

    See how it works

  • Reserved

    Capacity held for you

    GPUs or a cluster, priced per contract.

    Best for: Long training runsSteady inferenceMulti-node jobs

    Plan a reservation

GPU virtual machines

Launch a virtual machine with root access in our European data centres and install what you need. You pay per second while it runs.

GPU prices and capacity

US dollars per GPU hour, billed per second with no subscription. Current capacity is in your dashboard.

GPU and memoryPrice per GPU hour, spot and on demandLaunch
NVIDIA B300288 GBOn demand $7.99 per GPU hour, spot $2.99, 63% less.Launch
NVIDIA B200192 GBOn demand $6.49 per GPU hour, spot $2.40, 63% less.Launch
NVIDIA H200141 GBOn demand $4.29 per GPU hour, spot $1.60, 63% less.Launch
NVIDIA H10080 GBOn demand $2.79 per GPU hour, spot $1.10, 61% less.Launch
NVIDIA A10080 GBOn demand $1.59 per GPU hour, spot $0.80, 50% less.Launch
NVIDIA L40S48 GBOn demand $1.19 per GPU hour, no spot price.Launch

Size a model

See the memory a Hugging Face model needs and how many requests it serves.

Try

Context length

Tokens per request

Requests at once

Users or streams at the same time

Reserved GPUs and clusters

On-demand capacity depends on what’s free when you launch. Reserved GPUs stay yours, at a price agreed up front.

From a few GPUs to 1,000s. Start small and scale to thousands of GPUs as your workload grows.

See GPU blocks available now

Plan a reservation

Your answers pre-fill the booking form.

GPU
GPUs in total
Start
For how long
Region

Serverless training

Train models without managing machines: package your training code in a container, choose a GPU and submit the job.

  1. You

    Package your code

    Put your training code and its dependencies in a plain Docker container.

  2. You

    Choose a GPU and submit

    Pick the GPU your model needs, check its price, then submit the workload.

  3. We

    We run it on that GPU

    The job runs on the hardware you chose. When it finishes, you inspect its output.

train.sh
# 1  Package your code
$ docker build -t your-team/train .
$ docker push your-team/train# 2  Choose a GPU and submit
$ lyceum docker run your-team/train \
    -m gpu.h100 -c "python train.py"# 3  We run it on that GPU
#    The output streams here until
#    the run ends. Add -d to detach.
Example image and command · Lyceum CLI
Read the training docs

Pick a GPU and start building

Start buildingOpen dashboardTalk to an expert