When I ask the chat about something it doesn’t do anything, yet it uses some of my credit everytime.
I uploaded a video as an explanation: https://www.youtube.com/watch?v=COI6ou5vBWI
I’m having the same issue but only with claude 3.5 sonnet. More than half of my requests with that model do absolutely nothing except use my credits. Switching to gpt-4o solves for me.
I’m experiencing the same thing with Claude 3.5 sonnet. Was just about to upload a video. I can enter a prompt, but it returns nothing, and the premium count increments by 1 each time.
Same, on Pro account, most of my Sonnet request do nothing and display ‘completed’ status text.
Sorry about this! Seems like Anthropic is having some downtime so Sonnet requests might fail - GPT 4o should be working as expected!
Is it possible to add an IF conditional like:
IF request response status is not 200 (OK), then don’t increment request counter, but still add it to the request logs with its failed status.
A ‘request info indicator’ might be helpful in the UI too with info like:
- Model used
- Response status
- Response duration (so you can see from your end when things are slower than usual)
- Request token count
- Total chat token count (feature request)
It might help to avoid this ‘silently failing’ dynamic.
That’s what I was thinking too. Is it really hard to just check if the request went through and then use teh credit? I wasted around 10 credits thinkinging there was a bug with the editor itself not knowing it was using my credit
@rishabhy will the failed / empty fast requests going to be returned for this month’s quota?
would love to return the failed requests - please email us at hi@cursor.com and we’ll get it sorted out!