Describe the feature or problem you'd like to solve
Using the /model command to switch reasoning effort is multi-step and cumbersome.
I often want to adjust the reasoning effort of a prompt based on the complexity of the ask. For performance, I'll lower the reasoning effort to "Low" or "Medium" for straightforward requests. Then raise to "High" or "xHigh" when implementing features, design docs, multi-step workflows etc. I generally don't want to switch the actual model; I just want to adjust the reasoning effort.
The current experience for doing that using the /model command is fairly cumbersome. It's a multi-step process because the current model isn't highlighted by default; I must first navigate to the same model again, select it, then change the effort.
Proposed solution
Introduce an /effort command that allows you to switch reasoning effort on the same model in one command.
E.g.
/effort Medium or /effort med/effort + arrow keys to select, Enter
Example prompts or workflows
Desired:
- Using Opus 4.7 on xHigh, I plan and implement a new feature. Great!
- "Hmmm, I want this variable to have a different name"
/effort med"Rename this variable to X"/effort xhigh
Actual:
- Using Opus 4.7 on xHigh, I plan and implement a new feature. Great!
- "Hmmm, I want this variable to have a different name"
/model- Type "4.7"
- Use arrow keys to select Opus 4.7 again, Enter
- Use left/right arrow keys to select Medium, Enter
"Rename this variable to X"/model- Type "4.7"
- Use arrow keys to select Opus 4.7 again, Enter
- Use left/right arrow keys to select xHigh, Enter
Because it's so cumbersome today, I will often just execute the rename on xHigh which in my experience takes noticeably longer.
Additional context
I suppose another way to mitigate this would be: Have the currently selected model highlighted by default when you execute /model. That would allow me to just hit Enter and select the effort, eliminating much of the churn. However, I believe an /effort command to be the ideal solution here.
(FWIW: Claude Code CLI has this command 🙂)
Describe the feature or problem you'd like to solve
Using the
/modelcommand to switch reasoning effort is multi-step and cumbersome.I often want to adjust the reasoning effort of a prompt based on the complexity of the ask. For performance, I'll lower the reasoning effort to "Low" or "Medium" for straightforward requests. Then raise to "High" or "xHigh" when implementing features, design docs, multi-step workflows etc. I generally don't want to switch the actual model; I just want to adjust the reasoning effort.
The current experience for doing that using the
/modelcommand is fairly cumbersome. It's a multi-step process because the current model isn't highlighted by default; I must first navigate to the same model again, select it, then change the effort.Proposed solution
Introduce an
/effortcommand that allows you to switch reasoning effort on the same model in one command.E.g.
/effort Mediumor/effort med/effort+ arrow keys to select, EnterExample prompts or workflows
Desired:
/effort med"Rename this variable to X"/effort xhighActual:
/model"Rename this variable to X"/modelBecause it's so cumbersome today, I will often just execute the rename on xHigh which in my experience takes noticeably longer.
Additional context
I suppose another way to mitigate this would be: Have the currently selected model highlighted by default when you execute
/model. That would allow me to just hit Enter and select the effort, eliminating much of the churn. However, I believe an/effortcommand to be the ideal solution here.(FWIW: Claude Code CLI has this command 🙂)