Benchmarking

Benchmarking lets you run the same prompt across multiple models, isolate changes, and apply the best result.

Start a benchmark (TUI)

In the TUI prompt:

/benchmark start

This creates a parent session and child sessions for each model you select.

Common commands

/benchmark stop
/benchmark next
/benchmark prev
/benchmark apply
/benchmark undo

Shortcuts (default)

• Next session: ctrl+shift+right
• Previous session: ctrl+shift+left
• Apply changes: ctrl+alt+a
• Undo changes: ctrl+alt+u

Typical flow

1. Run /benchmark start and choose models
2. Switch children with /benchmark next / /benchmark prev
3. Apply a child’s changes with /benchmark apply
4. Undo with /benchmark undo
5. Exit with /benchmark stop