Modal logo
IntroductionCustom container images Defining ImagesPrivate registriesFast pull from registryGPUs and other resources GPU accelerationUsing CUDA on ModalReserving CPU and memoryScaling out Scaling outInput concurrencyBatch processingJob queuesDynamic batching (beta)Dicts and queuesScheduling and cron jobsDeployment Apps, Stubs, and entrypointsManaging deploymentsInvoking deployed functionsContinuous deploymentRunning untrusted code in FunctionsSecrets and environment variables SecretsEnvironment variablesWeb endpoints Web endpointsStreaming endpointsWeb endpoint URLsRequest timeoutsWebhook tokens (beta)Networking Tunnels (beta)Proxies (beta)Cluster networkingData sharing and storage Passing local dataVolumesStoring model weightsDataset ingestionCloud bucket mountsSandboxes SandboxesRunning commandsNetworking and securityFile accessSnapshotsPerformance Cold start performanceMemory Snapshot (beta)Geographic latencyReliability and robustness Failures and retriesPreemptionTimeoutsTroubleshootingSecurity and privacyIntegrations Using OIDC to authenticate with external servicesConnecting Modal to your Datadog accountConnecting Modal to your OpenTelemetry providerOkta SSOSlack notifications (beta)Other topics Modal 1.0 migration guideFile and project structureDeveloping and debuggingModal user account setupWorkspacesEnvironmentsJupyter notebooksAsynchronous API usageGlobal variablesRegion selectionContainer lifecycle hooksParametrized functionsS3 Gateway endpointsGPU Metrics

Jupyter notebooks

You can use the Modal client library in notebook environments like Jupyter! Just import modal and use as normal. You will likely need to use app.run to create an ephemeral app to run your functions:

# Cell 1

import modal

app = modal.App()

@app.function()
def my_function(x):
    ...

# Cell 2

with modal.enable_output():
    with app.run():
        my_function.remote(42)

Known issues

  • Interactive shell and interactive functions are not supported.

    These can only be run within a live terminal session, so they are not supported in notebooks.

  • Local and remote Python versions must match.

    When defining Modal Functions in a Jupyter notebook, the Function automatically has serialized=True set. This implies that the versions of Python and any third- party libraries used in your Modal container must match the version you have locally, so that the function can be deserialized remotely without errors.

If you encounter issues not documented above, try restarting the notebook kernel, as it may be in a broken state, which is common in notebook development.

If the issue persists, contact us in our Slack.

We are working on removing these known issues so that writing Modal applications in a notebook feels just like developing in regular Python modules and scripts.

Jupyter inside Modal

You can run Jupyter in Modal using the modal launch command. For example:

$ modal launch jupyter --gpu a10g

That will start a Jupyter instance with an A10G GPU attached. You’ll be able to access the app with via a Modal Tunnel URL. Jupyter will stop running whenever you stop Modal call in your terminal.

See --help for additional options.

Further examples