Scores a batch of records against one or more saved models and returns the results inline. Synchronous: the call blocks until every record is scored. Pass several configIds to apply several models in one pass.

Unknown configIds are skipped, not rejected. A well-formed id that matches no existing config is silently ignored and the run proceeds on the remaining models, so verify your ids against List Score Configs rather than trusting a 200 to mean every model was applied.

Supply configIds (required, 1 to 10 score config ids) and records (required, 1 to 25 records).

Returns 200 with per-record scores inline. Returns 400 if configIds or records are missing or empty, if either exceeds its limit, if a configId is malformed (not a valid integer id), or if none of the supplied configIds match an existing config. Returns 5xx if the run fails before any record can be scored; failures affecting individual records are reported per-record in the response.

This operation consumes credits for each record scored. Config operations are free — only a run is billed.