Clef or Clef-flash in front of a Sume agent: which tier to use

Use Clef-flash for high-volume, low-stakes gates, and Clef for decisions where a wrong answer starts expensive work. Cloudflare's launch post gives a median latency of 38.8 ms for Clef-flash and 209.3 ms for Clef. Price is not disclosed on the page that was read, so budget with your own measurements.

Pick by the cost of a wrong yes

A Sume Agent Completion takes far longer than either decision, so latency of the gate rarely matters against the run itself. What matters is how often the gate is wrong and what a wrong yes costs.

Measure on your own data

Whatever the tier, the run still needs generation_spend_cap_usd. Measure both tiers on a labeled set of your own tickets, and compare how many wrong yeses each lets through, before you choose.

Check the docs before you ship

Sume's limits and field names change faster than blog posts do. Read the linked docs pages for the current request fields before you ship, and send a dry_run or a low spend cap on your first real call.