Workflow Interruption in Auto Model Mode vs. Claude Sonnet 4

Hi Cursor Team,

I’d like to report a key workflow discrepancy in Auto Model Mode:

Core Issue:
When using Auto Model Mode after exceeding Pro tier limits:

  • Workflows constantly require manual “Continue” confirmations
  • Execution gets fragmented across multiple prompts
  • This never occurs when explicitly selecting Claude Sonnet 4

Expected Behavior:
Auto Model Mode should deliver the same seamless experience as manually selecting Sonnet 4:

  • Complete instructions in single execution
  • Zero unnecessary confirmations
  • Continuous workflow without interruptions

Impact:
The current Auto Model behavior significantly reduces productivity compared to direct Sonnet 4 usage, undermining Pro tier value.

Could Auto Model Mode be optimized to match Claude Sonnet 4’s uninterrupted execution standard? Thanks for addressing this!

1 Like

I find the auto mode mainly uses GPT 4.1, and this is classic 4.1 behavior