Run GPU jobs from your terminal.
Create instances, move files and run scripts with the Jarvislabs CLI. Use the Python SDK in your code, or give your coding agent the same GPU workflow.
Free CLI and Python SDK · Python 3.10+ · Compute and storage billed separately
uv tool install jarvislabs
jl setup
jl run train.py --gpu A100Get started
Install once. Connect your account.
Install the jarvislabs package, then run jl setup to configure your API key.
uv tool install jarvislabspip install jarvislabspipx install jarvislabsChoose one installation method. Create an API key in dashboard Settings. For automation, supply it through the JL_API_KEY environment variable.
Managed runs
Send a script to a GPU.
Use jl run to upload your code, prepare a Python environment, install requirements and stream logs back to your terminal.
# Create an instance for this run
jl run train.py --gpu A100
# Run on an instance you already have
jl run train.py --on <instance-id>Follow the output
Read your script’s logs from your terminal while the work runs on the GPU.
Pause after the run
Managed runs pause the instance they create when the run finishes. Retained storage continues to incur charges.
Keep direct access
Use jl ssh to connect when you want to inspect files or work interactively.
AI-native workflows
Give your coding agent GPU access.
Set up Jarvislabs skills for Claude Code, Cursor, Codex or OpenCode through jl setup.
Work in your editor
Let your agent use the CLI to launch compute and run code while you review its work.
Use the same commands
Agent workflows use the same CLI you can run yourself. Inspect commands, instance state and logs.
Automate with Python
Use the Python SDK to manage instances from scripts and applications.
Read the SDK docs↗Instance control
Manage the full instance lifecycle.
Check GPU availability, create an instance, connect over SSH, and pause or delete it when you are done.
jl gpus
jl create --gpu A100
jl ssh <instance-id>
jl pause <instance-id>Before you start
Common questions.
Practical details for choosing and using this product.
Does the CLI cost extra?
The CLI and Python SDK are free. You pay for GPU compute and storage on the resources you create. Pausing compute does not delete retained storage.
What do I need to install it?
Python 3.10 or later and a Jarvislabs account. Install with uv, pip or pipx, then run jl setup.
Can I use it in CI?
Yes. Supply JL_API_KEY through your CI secret environment and use --yes for commands that require confirmation. The CLI and SDK support non-interactive workflows.
How do I see available GPUs?
Run jl gpus to see live GPU availability, memory and hourly prices by region.