I'm trying to figure out what model was used when I select Auto

Discussions
1

I cannot figure out how to determine what AI model was chosen when I use Auto mode. I can’t find anything in the UI and when I ask it in a chat I get a response like this:

I’m Auto, Cursor’s agent router—not a single fixed public model name on my side. Which underlying model is used for a reply is determined by Cursor’s routing and your settings, not something I can read out from here.

I’d like to be able to know what model is doing the work if I need to. When the model does a decent enough job for the request, I don’t care. But when it’s a complete freaking moron I really want to know. I really only use Auto mode when 1. I’m giving it a task that should be pretty easy and 2. I am OK with scratching its garbage and trying again with a model of my choosing.

As an engineer, using a tool to work against my codebase, I think it’s my right to know what is churning on my code and – most importantly – isn’t doing a good job on it. I’m OK for now wasting some time, from time to time and calling it R&D, but this isn’t going to go on forever.

4

Hey @RayGraham!

One of the trade-offs of the fixed price for Auto is that the exact Model isn’t available.

If you need to be certain about the model used, you’ll need to pick a specific model rathre than Auto.

6

I guess the best “equivalent” to Auto is to use Composer 2 instead. :smiley:

7

Many times, you can actually see it in https://cursor.com/dashboard/usage
Even if the model says “Auto” the graph will say “Default” and the used models xD
So one step closer at least.