> ## Content Index
> Fetch the complete content index at: https://christophermoravec.com/llms.txt
> Use this file to discover other available public pages before exploring further.

# Claude Clue #20: Effort Is the New Thinking
- URL: https://christophermoravec.com/claude-clues/claude-clue-20-effort-is-the-new-thinking/
- Published: 2026-05-05T15:12:45.000Z
- Updated: 2026-05-05T15:12:45.000Z
- Description: If you've been using Claude Code for a while, you might remember manually toggling between thinking modes or typing 'ultrathink'.
- Author: Christopher Moravec
- Tags: Claude Clues

If you've been using Claude Code for a while, you might remember manually toggling between thinking modes or typing 'ultrathink' when you needed the model to really chew on a problem. Anthropic has layered a new control on top of all that: "effort" levels. Claude 4.6 introduced four (low, medium, high, max), and 4.7 just slotted in a fifth called "xhigh" — sitting between high and max for long-horizon coding and agentic work. Same model, different gears. The tradeoff is straightforward: lower effort means faster responses and fewer tokens burned, while higher effort means more time to plan, reason, and explore. Worth noting — ultrathink still works as a per-prompt nudge when you want extra reasoning on a single turn without changing your session default.

Here's what this looks like in practice: when you run \`/model\` in Claude Code, up/down arrows pick your model and left/right arrows adjust the effort slider. After some back-and-forth in March, the current defaults are xhigh on Opus 4.7 and high on Opus 4.6 / Sonnet 4.6\. So if you've got a gnarly refactoring task and you're not in a rush, crank it to max and let it think. If you're just renaming variables, dial it back and save the tokens. It's a nice bit of fine-tuning that lets you find your own sweet spot between cost, speed, and capability.

*No robots were harmed in the making of this clue.*

This one took all three of us, Roscoe, Christopher and Jaws!