Decision-1 GlobalStandard or DataZoneStandard for a Sume gate?

Use DataZoneStandard when the briefs you gate must stay inside a data zone and the type is available in your region; use GlobalStandard when you want global capacity and can accept more variable latency. Microsoft says both are supported for Microsoft-Decision-1, and also says to benchmark because a deployment type does not guarantee lower latency.

What the page states

The numbers here are only what the Learn page says. It gives no price and no latency figure for either type.

Why it matters for a render gate

The gate sits on the path between a user pressing a button and a Sume video submit. That submit answers 202 and the work is asynchronous, so a slow gate is the visible part of the wait. Measure it on your own briefs with a handful of labelled examples, as Microsoft's quickstart does, and keep a short timeout that fails to review.

Also create the Foundry resource in a region near your application; the page recommends this to cut network latency.

What it does not decide

The deployment type does not affect spend on the Sume side. Sume's own controls (dry_run, max_spend_credits, idempotency_key) are described on the MCP tools and gates page.

How to compare on your own briefs

Create two deployments, one of each type, with the same capacity setting, then run the same twenty labelled briefs through both and compare median latency and any difference in answers. Microsoft's quickstart does the same with accuracy and median request latency, so you can adapt its script.

Choose on evidence: if the difference is small, pick on data-residency grounds.

• Same briefs, same questions.
• Compare accuracy as well as time.
• Re-run after a region change.