Gemini managed agents background:true vs a Sume Format 202 receipt
Both return an id to poll instead of holding the connection. Compare the Gemini background flag with the Sume Format run receipt and its four URLs.
Both are the same pattern: start long work, get an id back right away, and poll for the result. Google's July 7 post on the Gemini API's managed agents says that with background: true the call returns an id you can poll instead of keeping a connection open for a long task (read 2026-10-04). A Sume Format run does the equivalent by default: POST /v1/formats/{handle}/{slug}/runs answers 202 with a receipt.
If you already wrote a poller for one, most of it carries over to the other. The differences are in what the receipt hands you.
What does the Sume receipt contain?
Sume's Create a run page says the receipt carries the run id and four addresses, so you never build a URL by hand.
| Field | Use |
|---|---|
status_url | Poll for status; the run is terminal when status is completed, failed, canceled or skipped |
result_url | Fetch the result once the run is terminal |
events_url | Read the run's event history |
cancel_url | Cancel while the run is in flight |
How do I poll without hammering it?
Follow the next_action hint on the status response. Sume's Runs and results page defines three values: poll_status means check again, retry_later means a transient gap, and none means stop. Back off between polls, and set a ceiling: a run is force-finalized as failed 90 minutes after creation, so a loop that outlives that is a bug in your loop.
Do I need to poll at all?
Not if you can receive a webhook. Register a public HTTPS webhook_url on the create call and Sume posts a format.run.terminal event when the run ends, signed so you can verify it. Keep the poller as a backstop for a missed delivery, and call the redeliver endpoint when you need a replay.
What does the Gemini post leave open?
It says a background call returns an id to poll; it does not, in the part read, list the terminal states, a cancel call or a hard ceiling. Check Google's reference for those before you reuse a Sume-shaped loop against it.
Sources
Related posts
More in Agents
- Grok 4.7 Fast is only in Cursor and Grok Build, at 2x the rate
xAI's pricing page says Grok 4.7 Fast is not on the public API. Where it runs, what it costs, and what to do if your Sume agent needs a Grok model.
- Hermes cron fixes for run now, and Sume's one-active-run rule
Hermes fixed run-now cancelling the next scheduled run. Sume schedules allow one active run: a second is skipped, or rejected if you ask for reject.
- job_scope_forbidden: why an agent run cannot list every Sume job
A thread-bound Sume credential lists only its own thread's jobs, and unattended runs cannot ask for the whole workspace. What to use instead.
- Kilo scheduled sessions vs a Sume schedule for recurring renders
Kilo v7.8.3 reports a session waiting on a wakeup or cron task as scheduled. A Sume schedule lives server-side; use it when renders must run unattended.
Written by Sume