Cloud Notebooks

Jupyter on cloud GPUs. One command.

Launch a GPU-powered notebook environment in seconds. No infrastructure, no setup, no waiting.

Request Access View Docs
nb-h100-a7f2c1.lyceum.technology
Jupyter training_experiment.ipynb
H100 80GB Connected
Python 3.11 PyTorch 2.2
[1]:
import torch
import torch.nn as nn
from transformers import AutoModel
[2]:
torch.cuda.get_device_name(0)
'NVIDIA H100 80GB HBM3'
[3]:
model = AutoModel.from_pretrained("meta-llama/Llama-3.3-70B")
model.to("cuda")
Loading model weights... 68%

From terminal to notebook in seconds

No VMs to configure. No Jupyter to install. Just one command.

lyceum
notebook
launch

Run the command

Pick your GPU and image. That's it.

GPU

GPU provisioned

We spin up your machine with Jupyter ready.

Jupyter

Start coding

Jupyter opens in your browser automatically.

Flexible Hardware

Any GPU, any time

From cost-effective T4s for exploration to H100s for serious training. Switch GPUs between sessions as your needs change.

T4 L40S A100 H100 H200
T4
16 GB VRAM
$0.39/hr
A100
80 GB VRAM
$1.39/hr
H100
80 GB VRAM
$2.49/hr
Terminal
# Use a pre-built image
$ lyceum notebook launch
# Or bring your own
$ lyceum notebook launch -i myregistry/ml-env:v2
# With a GPU
$ lyceum notebook launch -m h100
Your Environment

Bring your own stack

Use pre-built images with PyTorch, JAX, or TensorFlow. Or bring your own Docker image with all your dependencies pre-installed.

  • Pre-built images for common frameworks
  • Custom Docker images from any registry
Pay for What You Use

Per-second billing, auto-shutdown

Set a timeout and your notebook automatically stops when idle. No surprise bills from forgotten instances.

Session Cost A100 80GB
Active time 47m 23s
Rate $1.39/hr
Total $1.10

Simple, per-second pricing

Pay only for the time your notebook is running. Stop anytime.

GPU
VRAM
Price
NVIDIA B200
192 GB
$5.89/hr
NVIDIA H200
141 GB
$3.69/hr
NVIDIA H100
80 GB
$3.29/hr
NVIDIA A100
80 GB
$1.99/hr
NVIDIA L40S
48 GB
$1.49/hr
NVIDIA T4
16 GB
$0.39/hr

View all GPU options →

Start exploring in seconds.

Launch a GPU-powered notebook with one command.

Request Access View Documentation
Book a Demo