Combines one exclusive Choice category with independent Noul tags, runs a bounded-concurrency batch, selects tag thresholds on labeled tweets, and calculates cost from returned usage.
Original by OpenRouterClassificationOpenRouter cookbookSource reviewed
Before you dive in
What you’ll find in the original
Use Choice for one category and Nouls when several tags can apply.
Choose thresholds on a separate human-labeled sample.
Bound concurrency and record usage so cost per thousand items reflects real requests.
Worth knowing
Its TweetTopic calibration and batch are a worked example, not proof that the same thresholds transfer to your data.