Back to AIBriefs
AnalysisAI Models

Reddit user notes Opus 4.8 effort scaling changes

A Reddit user reports that Opus 4.8's "low" effort now uses as many output tokens as medium-high effort on Opus 4.7 or 4.6. The "medium" effort setting consumes more tokens than 4.7 high or nearly as many as 4.6 max.

·
May 29, 1:29 PM
Reddit user notes Opus 4.8 effort scaling changes — AIBriefs