Effort Level
A core parameter in Claude Code that controls the depth of model reasoning and token consumption. It is divided into multiple tiers ranging from low to high (e.g., low, medium, high, xHigh). Higher tiers cause the model to allocate more thinking tokens, yielding better performance on complex tasks at greater cost.
Timeline (last 90 days)
Lower effort levels can alleviate the model's rabbit-holing behavior, but setting it too low may also cause the model to give up on a problem prematurely.
The effort parameter controls how much the model thinks rather than how much it says; reducing effort does not reliably shorten visible responses, and controlling response length requires explicit prompting.
The effort level is the primary means of controlling token cost and response time; the official recommendation is to start with the default high intensity and adjust based on evaluation.
All Facts (3)
The effort parameter controls how much the model thinks rather than how much it says; reducing effort does not reliably shorten visible responses, and controlling response length requires explicit prompting.
60%UnverifiedLower effort levels can alleviate the model's rabbit-holing behavior, but setting it too low may also cause the model to give up on a problem prematurely.
50%UnverifiedThe effort level is the primary means of controlling token cost and response time; the official recommendation is to start with the default high intensity and adjust based on evaluation.
50%