GPT-4.1 is now available in Cursor

You can now enable GPT-4.1 in Cursor from Settings → Models.

It’s free for the time being to let people get a feel for it!

We’re keeping a close eye on tool calling abilities and will be sharing feedback with the OpenAI team as things evolve.

Try it out and let us know what you think!

48 Likes

Awesome! Thanks!

1 Like

That was fast! Thanks :+1:

1 Like

image
Ask him if he wrote this wrong, and whether his context is 120K or 60000.

sweet

updated both pricing and context window size!

1 Like

great

its fast …

I need to know something, I have a Pro account with 1,000 fast requests. I would also like to enable usage based pricing, but not if my account reverts back to only 500 fast requests. What does enabling usage based pricing do in situations like mine ?

I’m almost positive it doesn’t revert your fast requests. All usage based pricing does is charge you per-request after you’ve used up all your fast requests (including bought requests).

Fast request packages won’t be supported going forward, so would recommend downgrading. Pricing is the same ($20/500 = $0.04 per request)

I have a question,
Until when is it free? 1 month ? end of april ? or end of 2025? What is the plan ?

Thanks ! Cursor !

2 Likes

Hey, awesome job adding GPT-4.1 so quickly. Really impressed by your speed

1 Like

One thing I’ve noticed right away with GPT-4.1 is that it follows your defined rules—like AI behavior rules or Cursor project rules—far more strictly than models like Claude or others.

This means you need to be much more intentional with how you write your rules for GPT-4.1. The old scattershot approach (where you’d throw a bunch of rules in and hope Claude would latch onto one) might backfire with 4.1, it will try to follow all of them—so redundancy or conflicting instructions can cause repetition or confusion.

9 Likes

This is really frustrating … that GPT 4.1’s 1 million token context is hamstrung by Cursor imposing 128k … if that. Ugh. Feels disgusting.

3 Likes

Is that true @cursor-team ?

2 Likes

Access blocked, please contact your admin

Others are offering ‘Unlimited GPT-4.1 for free from April 14 to April 21’, but for some reason, Cursor AI decided to reduce this model’s capabilities by 10 times — from 1 million tokens to 128k.

What’s the reason for limiting a free model that is specifically designed for a large context?

7 Likes

I believe cause its still expensive to run, they might release a max with the 1 mil, which makes more sense with the infra

Hello,
I was able to test GPT 4.1 but I don’t understand some things. I thought Grok was going to revolutionize certain things, now GPT4.1 etc… And nothing, I have the impression of using the same AI everywhere.
Is it a software context issue that’s blocking?
There must be a problem!!! The new GPT is hyper good in coding but we don’t see any benefit on the agent mode.

Please, we want answers

1 Like