Yes — in Prefect 3.x, that combination behaves exactly as you described: - With

PREFECT_TASKS_DEFAULT_PERSIST_RESULT=false

, task results are not persisted by default. This task-level default takes precedence for tasks. - With

PREFECT_RESULTS_PERSIST_BY_DEFAULT=true

, flow results are persisted by default. There’s no separate flow-level default env var, so flows use this setting. You can still override per object: - Tasks:

@task(persist_result=True/False)

- Flows (including subflows):

@flow(persist_result=True/False)

Example:

from prefect import flow, task

@task(persist_result=True)  # overrides the task default
def compute(x):
    return x + 1

@flow(persist_result=False)  # overrides the flow default if you don't want the final flow result persisted
def my_flow():
    return compute(1)

Docs and references: - Results and persistence overview: Results (Prefect 3 docs) - Relevant source (for precedence and defaults): - prefect/results.py - prefect/context.py - prefect/flow_engine.py Note: Ensure you have appropriate result storage configured (e.g., via

result_storage

on a flow/deployment) if you want persisted results to be written to a durable location. Subflows are flows, so they’ll follow the flow rule above unless you override

persist_result

on the subflow decorator.