Claude seems both stupider and more expensive than before

A user reports that Claude, specifically Opus and Sonnet subagents, seems both "stupider and more expensive" than before. The user's workflow involves a Fable orchestrator managing these subagents for tasks, including code review. They note that previously, the Opus subagent would have been more efficient, and the Fable orchestrator would have adapted its approach or sought human intervention when encountering issues. The user also suggests that their quota usage indicates a change in performance, as they previously couldn't consume it so quickly with the same number of agents.

This report details a user's direct experience with Claude's perceived degradation, unlike many other reports that focus on general observations or benchmark changes.