Modal logo
IntroductionCustom container images Defining ImagesUsing existing container imagesFast pull from registryGPUs and other resources GPU accelerationUsing CUDA on ModalReserving CPU and memoryScaling out Scaling outInput concurrencyBatch processingJob queuesDynamic batching (beta)Multi-node clusters (beta)Deployment Apps, Functions, and entrypointsManaging deploymentsInvoking deployed functionsContinuous deploymentRunning untrusted code in FunctionsModal Sandboxes SandboxesRunning commandsNetworking and securityFile accessSnapshotsModal NotebooksSecrets and environment variables SecretsEnvironment variablesScheduling and cron jobsWeb endpoints Web endpointsStreaming endpointsWeb endpoint URLsRequest timeoutsProxy Auth TokensNetworking TunnelsProxies (beta)Cluster networkingData sharing and storage Passing local dataVolumesStoring model weightsCloud bucket mountsDictsQueuesDataset ingestionPerformance Cold start performanceMemory SnapshotHigh-performance LLM inferenceGeographic latencyReliability and robustness Failures and retriesPreemptionTimeoutsGPU healthTroubleshootingSecurity and privacyIntegrations Using OIDC to authenticate with external servicesConnecting Modal to your Datadog accountConnecting Modal to your OpenTelemetry providerOkta SSOCustom SAML SSOSlack notifications (beta)Workspace & account settings WorkspacesEnvironmentsModal user account setupService usersBillingOther topics JavaScript/Go SDKsModal 1.0 migration guideFile and project structureDeveloping and debuggingDeveloping Modal code with LLMsJupyter notebooksAsynchronous API usageGlobal variablesRegion selectionContainer lifecycle hooksParametrized functionsS3 Gateway endpointsGPU Metrics

Parametrized functions

A single Modal Function can be parametrized by a set of arguments, so that each unique combination of arguments will behave like an individual Modal Function with its own auto-scaling and lifecycle logic.

For example, you might want to have a separate pool of containers for each unique user that invokes your Function. In this scenario, you would parametrize your Function by a user ID.

To parametrize a Modal Function, you need to use Modal’s class syntax and the @app.cls decorator. Specifically, you’ll need to:

  1. Convert your function to a method by making it a member of a class.
  2. Decorate the class with @app.cls(...) with the same arguments you previously had for @app.function(...) or your web endpoint decorator.
  3. If you previously used the @app.function() decorator on your function, replace it with @modal.method().
  4. Define dataclass-style, type-annotated instance attributes with modal.parameter() and optionally set default values:

The parameters create a keyword-only constructor for your class, and the methods can be called as follows:

Function calls for each unique combination of values for foo and bar will run in their own separate container pools. If you re-constructed a MyClass with the same arguments in a different context, the calls to baz would be routed to the same set of containers as before.

Some things to note:

  • The total size of the arguments is limited to 16 KiB.
  • Modal classes can still annotate types of regular class attributes, which are independent of parametrization, by either omitting = modal.parameter() or using = modal.parameter(init=False) to satisfy type checkers.
  • The support types are these primitives: str, int, bool, and bytes.
  • The legacy __init__ constructor method is being removed, see the 1.0 migration for details.

Looking up a parametrized function 

If you want to call your parametrized function from a Python script running anywhere, you can use Cls.lookup:

Parametrized web endpoints 

Modal web endpoints can also be parametrized:

Parameters are specified in the URL as query parameter values.

Using parametrized functions with lifecycle functions 

Parametrized functions can be used with lifecycle functions.

For example, here is how you might parametrize the @modal.enter lifecycle function to load a specific model:

Performance 

Currently, parametrized Function creation is rate limited to 1 per second, with the ability to burst to 1000. Please get in touch if you need higher rate limits.