Short answer: not directly. In Prefect 3.x,

refresh_cache

is a task-level setting; there’s no flow-level switch to “retry the flow with refresh_cache=True.” Ways to get the effect on a flow retry: - Set it on the tasks that should refresh on retry

from prefect import flow, task

  @task(retries=3, refresh_cache=True)
  def do_work(x):
      ...

  @flow(retries=2)
  def my_flow(x):
      return do_work(x)

- Turn it on globally for the run/deployment via settings/env var

# e.g., in your worker/deployment environment
  export PREFECT_TASKS_REFRESH_CACHE=true

Docs: Settings reference (see

tasks.refresh_cache

/

PREFECT_TASKS_REFRESH_CACHE

) - Make cache keys change on retries so cached results aren’t reused - Include the flow run attempt in your task’s

cache_key_fn

to force a new key on each flow retry:

import hashlib, json
  from prefect import task
  from prefect.runtime import flow_run

  def key_with_retry(ctx, task_args):
      attempt = flow_run.run_count() or 0
      args_hash = hashlib.sha1(json.dumps(task_args, sort_keys=True).encode()).hexdigest()
      return f"{ctx.task.name}:{attempt}:{args_hash}"

  @task(cache_key_fn=key_with_retry)
  def do_work(x):
      ...

Docs: Caching concepts - Disable caching on tasks you always want to recompute on retry

from prefect import task
  from prefect.cache_policies import NO_CACHE

  @task(cache_policy=NO_CACHE, retries=3)
  def do_work(x):
      ...

Notes - The

@flow

decorator has

retries

but no

refresh_cache

parameter. - There isn’t a CLI/UI toggle to “retry with refresh_cache” at the flow level. - Advanced: you can use API state transitions with

state_details.refresh_cache=True

, but that’s generally only for custom orchestration. Docs: Set flow run state. If you can share whether you want to recompute everything on retry or only specific steps, I can suggest the minimal change for your flow.