Which compute option should be chosen in a scenario where small-scale ad hoc Python scripts...

For small-scale, ad hoc Python workloads that run frequently and should start quickly and stop without leaving idle infrastructure running, serverless compute is the best fit. Databricks documentation explains that serverless compute allows users to run workloads without provisioning compute resources manually , and Databricks automatically allocates and manages the required resources. This improves startup and scaling speed and minimizes idle time, which directly matches the requirement that compute should “wind down quickly” after the work completes. Databricks also recommends serverless compute broadly when the workload is supported because it is the simplest and most reliable option. An all-purpose cluster is better suited to interactive collaborative development and can remain running longer than needed. A job cluster is ephemeral and often a good batch choice, but it still requires more explicit configuration than serverless and is not as optimized for minimal operational overhead. A SQL warehouse is intended for SQL workloads, not general ad hoc Python script execution. Therefore, serverless compute is the most appropriate choice.