โšก

Effort Levels

Claude Code

Token usage vs intelligence level control

Effort levels control the amount of tokens Claude invests in a task. Use Low for simple questions, Medium for general coding, and High for complex architecture design or debugging.

How to Use

1

Run /model command

2

Low: Minimum token usage, fast response โ€” suitable for simple questions or file checks

3

Medium: Balanced operation โ€” suitable for general coding tasks

4

High: Maximum token usage, highest intelligence โ€” suitable for complex problem solving

Tags

#model #token #performance #cost

Source

Boris Cherny (@bcherny)

← ๐Ÿ–ฅ๏ธ Terminal Setup ๐Ÿš€ Permanently Set Effort Level to Max →