[Claude Code Caught in 'Dumbing Down' Controversy Again: High Effort Selected but Only Shows 10, Official Says It's Just a Scale Change] Beating AI Newsflash: Claude Fable 5 is once again suspected by users of being secretly 'dumbed down.' Developers discovered that even though 'high effort' was selected in Claude Code, the model reported its effort value as only 10. Since 10 previously corresponded to 'low,' suspicions arose that Anthropic had secretly downgraded 'high' to what used to be 'low.' Claude Code team member Thariq later denied this. He explained that Anthropic is testing different backend configurations, and the numerical mapping for effort has also changed. The value of 10 does not mean 'only 10 out of a maximum of 100'; the number alone is meaningless. When users select 'high,' they are still receiving 'high' effort in practice. It's not surprising that users are upset. Back in March, Anthropic had lowered Claude Code's default effort from 'high' to 'medium.' After users complained that the model had become less intelligent, the setting was reverted. Anthropic later admitted that this was a mistaken trade-off. [Original Article Link]