GPT 4.1 is pretty good and fast! Quality-wise, it’s better than 4o but not as good as Sonnet + Gemini. It’s a good replacement for daily tasks. But Cursor charges 1 fast request for it, so why use it instead of 3.7?
Personally, I’m hoping for a model that balances quality and price. With 4.1 officially priced by OpenAI at 1.1 input and 4.4 output, I was hoping Cursor would price it at 2-3 requests/1 fast credit like o3-mini.
That price is for o4-mini(and the cached input is priced at $0.275). GPT 4.1 is priced at $2.00 input, $0.50 cached input, $8.00 output. This is pretty close to Sonnet and Gemini 2.5 pro.
Although I agree that o4-mini should be priced at 1/3 fast requests, like o3-mini is. And actual o3 should be at $0.25 per request, instead of $0.30. With the improvements in cached input costs(25% of the normal input in o3 and o4-mini compared to 50% in o1 and o3-mini), that would be a closer representation of the o3 costs compared to o1’s $0.40 per request.
Don’t you forum censors think that your silencing attempts will somehow stop people about trying out cheaper and/or better services. People will just be redirected by community to more open spaces like reddit, discord, x.
I have verifed the highlighted parts in an official blog post of the water competitor. They will price both models - 4.1 and o4-mini (though might be only medium) - as quarter of a use (1/4, 0.25 of premium model use). You and them both give 500 credits per month, so it is quite apples to apples comparison. Even if it is only o4-mini-medium, I would still rate that better, because your o4-mini high is not four times better as your pricing suggests. I really don’t understand why you are blocking stating your competitor’s name or their services.
You should respond with better or at least different product, not with silencing and banning users who are making comparisons. That only makes you look weak .
AI Models in Cursor (the table with cursor models, pricing api and cursor, aider rating)
Instead of 1/2 per request, they don’t want to add a price option. They’re choosing 1 credit instead of adding a 0.5. Maybe 1 and 0.333333333333333333333 are their lucky numbers!
As far as I know, they didn’t really explained the reasoning why o4-mini-high is raising price. Looking at data, o4-mini-high is actually like 25% cheaper in API, because of lower thinking coef (from Dubesor’s bench).
.
