Has anyone noticed that Cursor’s ‘Auto’ mode actually counts tokens? This means we’re forced to burn through our Claude model’s token quota before ‘Auto’ even kicks in.
cc @deanrie
From what I was told this doesn’t actually count against your premium usage but who knows with how everyone seems to be having the same issue of burning through tokens at an accelerated rate.
Auto usage doesn’t count against user selected premium model limits, so you can use auto without burning through Sonnet usage.
4 Likes
@Zackh1998 Same for me, getting charged for using auto mode.
I got an answer from support that only the ‘Cost to you’ column is important to us. So when you are using ‘auto’ it should not increase, the api cost is only an information for us.
Here is there answer:
Hi there! Let me clarify how the Auto model works with your usage.
Looking at your dashboard, while you see token counts and API costs for Auto model usage, these are just informational. On your Pro plan, Auto model usage is completely unlimited and does not count against your monthly usage limits. The "Cost to You" column correctly shows $0 for all Auto usage.
You can continue using the Auto model as much as you'd like without impacting your monthly $20 usage budget. This budget only applies when you manually select specific models.
You can read more about our models and pricing here: https://docs.cursor.com/models
I think it’s a bit confusing why all of a sudden some requests are shown in there, but others are not. Kind of random ![]()
2 Likes

