Hey everyone, I’m new to Cursor, and the past few days have been amazing! I’m not a developer, but I’ve managed to build a lot of things on my own. I was using only Claude, which was working well until I got this message: “Anthropic is experiencing a partial outage right now. To avoid downtime, we’ve temporarily fallen back to an OpenAI model.”
So, I tried using the other models available, but honestly, they don’t compare—even ChatGPT performs better! As a non-developer, it’s hard for me to work efficiently with these other options. Can anyone explain why the quality drops so much with other models? Is it the models themselves, or is it an issue with Cursor?
It probably defaults back to GPT 4o which I’ve found to be ass for basically all coding tasks. When there’s an outage I switch to gpt-4-turbo-2024-04-09 which is at least comparable to claude 3.5
It’s weird. I haven’t yet understood why using gpt-4o in cursor seems to underperform quite so markedly compared to taking the code and relevant discussion and pasting into a ChatGPT session – also configured for 4o. I guess it must be something about the way the context is condensed, and the fact that starting a fresh conversation with ChatGPT forces you to include only the relevant parts, less diluted? But I agree it does seem like there’s a really heavy penalty to the quality with anything other than the Anthropic models.