How to Quickly Use Claude 4 on Cursor for Advanced AI Tasks

I never imagined that Cursor would integrate Claude 4 so quickly after its release—it’s moving at lightning speed!

What’s more, using Claude 4 in Cursor comes with a limited-time free offer:
– A single use of “claude-4-sonnet” only costs 0.5 requests, making it 50% cheaper than “claude-3.7-sonnet.”
– Using “claude-4-sonnet (thinking)” once consumes just 0.75 requests, which is an impressive 62.5% discount compared to “claude-3.7-sonnet (thinking).”
While Claude 4 isn’t exactly budget-friendly, it seems like Cursor is pulling out all the stops to entice users.

Interestingly, when Cursor introduced Claude 4, they removed Claude 3.7 as a default option but retained Claude 3.5. Could it be that Cursor senses some inconsistency in Claude 3.7’s programming capabilities? Perhaps Claude 3.5 remains the safer and more reliable choice for now.

Has anyone experimented with writing code using Claude 4 yet?