Is gpt-4.1-mini via the Responses API better at argument setting or tool selection (between two similar tools).
-
Here, “Tool selection” is when the model must decide which function to call. Example: if its better with tool selection, then have two separate tools instead of a single one that uses a boolean arg to differentiate behavior.
-
“Argument setting” is when the model must decide how to populate parameters. Example: combine two tools into a single one, using a boolean arg to set behavior.
Have you had any experience with this?