Thx for your great work! I am really looking forward for the new model capabilities, since I am currently using gpt3.5 with the openai api key. With the larger token length and if I like the new performance, I am happy to finally upgrade to cursor pro ^^
they clearly couldn’t have done DevDay without Cursor! (& they should appreciate this with a better rate limit.)
would you consider a feature request for an initial rollout to folks who are BYOK?
(edit: if we promise not to complain when it doesn’t work as well as “old” GPT-4 )
Are you able to set any expectations for us if the plan is to go above the current 8k token size to utilise the new 128k potential?
I love cursor, but honestly the token limit is the biggest down side for me. Eats up so many hours having to chop up components into a state where the limited context would get a decent answer.
Am a bit worried since the previous stance has been not to go from 8k to 32 when it was an option. With the new 128k option is the plan to ramp this up?
I think many cursor customers would appreciate some clarity on this.
Related follow up with respect to pricing. Given that GPT4-turbo API price is much lower. Can we expect the number of queries in the paid plan to increase?
I also prefer if they increase the token size , the more history the chat has the dumber it becomes currently
Hello. Your “sad update” was posted a week ago. Initially, I thought it would occur within a few days and decided to stick with Cursor, waiting for this because someone had previously mentioned it would be “very soon,” which sounded like it would be a matter of days. However, a week has passed, and I feel compelled to ask for any clarification about your plans. Will it be days? Weeks? Months? As long as there is no news or estimates, I’m considering purchasing a ChatGPT subscription (specifically for coding), and it would be disappointing to learn that you’ll implement the new API the next day… I’m not asking you to expedite the process if it’s not feasible, but could you please provide any estimates of the timeline? I believe I’m not the only one who would appreciate this information.
GPT4-Turbo works now if you update cursor and turn on your API key. Note this obviously means you pay for it (to openAI). They mentioned they are testing it this way before rolling to paid plans.
Feels like it has a bit more “memory” from those increased tokens, but hard to say to honest. Would be nice to get some solid clarify from the Cursor team on all this.
Hey! Sorry about this.
It’s hard to give predictions here, because we’re blocked on OpenAI putting the model on our dedicated capacity offering (we don’t have high enough rate limits to use the public API for turbo).
We’re still very unsure when exactly this will happen, but given what OpenAI is telling us, it feels likely that it’s within a week.
That’s not working for me, and I’m not sure why. I’m running Cursor 0.15.4 on MacOS (Intel chip), using an OpenAI key. In the dropdown, I only see options for gpt-4 and gpt-3.5.
When I go to OpenAI’s playground, I do see an option for gpt-4-1106-preview. Not sure why it’s not showing up in Cursor. Anyone have any ideas?
update cursor? make sure it’s set to “use your key”?
Yeah, both of those are in place.
Does it work if you try turning your key on and off?
Ive found you need to:
- update cursor
- turn on key
- turn off cursor entirely
- turn cursor back on
Then you see the option
I updated Cursor again today (to 0.15.5) and the option is there now. Hooray!
guys who use the gpt4-turbo version, is it much better than gpt4?
How to make an update Cursor? Can you guide me…
In theory, you should be able to press Ctrl/Cmd+Shift+P and run the command “Cursor: Attempt Update”. In practice, this doesn’t always work, and you may need to redownload Cursor.
Any updates 20 days later?
I believe this is in place now. For me, Cursor 0.15.5 defaults to the “gpt-4-1106-preview” model.