The free stuff from OpenAI? I’m kinda skeptical about what it can do. 
Just click the “+ Add model” button, enter the name gpt-4.1, and press Enter.
What’s the best practice for using models from the same family?
More specifically, if I select GPT-4.1, should I deselect all other OpenAI models? Or is there a reason to keep multiple models from the same provider selected?
@deanrie Has there been any announcement on 4.1 being free indefinitely, the models page has dropped the end date and it just says free.
It too did not work. It looks like when I updated Cursor, it did not update the settings file or something, and even if I tried to reinstall it, the problem was still there.
The way I managed to fix it is to uninstall Cursor and delete .cursor in the user folder, and in AppData/Roaming and AppData/Local. Only then I got a fresh install that is working now.
Yes it was there, but it disappeared. I’ll try to find out.
the new models are working great for me in the chat, but when i use the edit (Cmd + K) its blank output?
im using a free cursor account with my openai api key plugged in.
edit: anyone else experiencing this?
To clarify, 1M tokens would cost $2 per request as per OpenAI documentation.
3 Likes
just use alipay
yea we need GPT-4.1 mini, 4.1 is maybe not ideal price wise.. lol
Wondering: is the fine-tuning option offered by OpenAI something that Cursor would want to invest a look into?
extremely frustrating behavior indeed. that seems to be one of the advantages of using claude in agent mode. GPT 4.1 acts like it is in “ask mode” and spends more time time explaining what the developer needs to do, instead of getting things done. All that asking amounts to wasted requests. TBH I wouldn’t use 4.1 in Cursor if it wasnt free.
2 Likes
I don’t like chatgpt in general. It’s like a senior dev who got bored of doing changes and fixing issues. Just sitting around and telling others what needs to be done.
Even for 4.1 they couldn’t solve this problem. Always unnecessarily reluctant.
Fantastic, 
!!!
4.1 does terrible for coding though, benchmarks are also disappointing. Seems that Gemini might have taken Claude’s spot though in the meantime, which is surprising.
2 Likes
great!
you are right, i am comparing with claude sonnet 3.7, sonnet is better than 4.1
As i calculated last 3 days for FREE 4.1 GPT.
Because GPT 4.1 is free, I can achieve my goal by drawing cards multiple times. Currently, the effect is acceptable to me.
it is costed normally and it is as expensive like sonnet 3.7 , so i cant understand why gpt can not create better models with the same price.