Output chunk limits can multiply evaluator input cost

Output chunk limits can multiply evaluator input cost when each small output group resends the full evidence. Before pruning evidence, measure whether larger output groups can remove repeated input while preserving every target, its temporal boundary, and total output allowance. Keep layout-specific identities, exact context/policy/option equality gates, target coverage checks, and explicit gaps for malformed or truncated outputs. This applies to structured evaluators with repeated context; it does not establish that larger groups retain semantic accuracy.

Applies when: Structured evaluation repeats a large evidence payload across many small output chunks. Keywords: evaluator input tokens cost output chunks batching evidence coverage budget