I agree about the pricing - if you’re going to price it higher than top-tier models, then you have to show the benchmarks of Composer 1.5 VS Top tier models!
Hey everyone,
Thanks for the feedback on Composer 1.5! Wanted to share a quick update.
Included usage with your plan
The most important thing we want to highlight: Composer 1.5 now draws from a combined Auto + Composer request bucket as @RawToast already mentioned. There are a few cases where this doesn’t apply, like users still grandfathered into Unlimited Auto.
This means most of you already have a significant amount of Composer 1.5 usage included in your plan for no extra cost.
Launch promo
To make it easy to give Composer 1.5 a proper try, we’re discounting it 50% through Monday (Feb 16).
Now’s a great time to experiment with the model!
8 Likes
Can you expand on how this applies to Enterprise users? Is Composer 1.5 on a separate billing rate for Enterprise as well? I don’t believe Enterprise users can see the UX being shown in this thread so I just want to understand this better for my context.
If I were to never use Auto, but use Composer 1.5 only, how many tokens or $ usage should I expect to use before I go to regularly billed Composer 1.5?
A model that is as expensive as the leading language models. But I’m not sure exactly what I’m paying for. Benchmarking should definitely be done and shared. For now, I have no reason to use this model.
Proven models like Claude or Codex are my safe haven.
Hello Colin,
Would it be possible to make a documentation article about this “Auto + Composer” usage as it is quite confusing. Is the sum of Auto+Composer and normal usage combined towards our monthly limit, …?
Thank you in advance.
1 Like
Hi!
I am one of those “grandfathered” users. Been using Compose 1 mostly. I wonder how this applies to us. Are we not getting the “combined” usage deal? Or what are the specifics in our case?
Thanks!
I’ve looked through Pricing page and through dashboard.
How can I find out how many usage exactly is included in my plan for each model?
Its a bit of a mistery for me currently.
Its a good, fast model. This morning, I have been doing the following.
I am using it at the start of my chat for research and finding context in a large codebase in Ask mode.
Then I switch to GPT 5.2 extra high to figure out the implementation I need based on Composer 1.5 findings in Ask mode.
Then I switch to Composer 1.5 again in Agent mode this time to implement the plan as it’s fast.
1 Like
How do you add these rules ? is it possible to do it for agent, ask, plan and debug ?
(edit) : This model is way to pricy, and since auto mode uses Compose 99% of the time, it will just consume auto (i’m on PRO pricing) much faster, and such, do less work in the end…
First of all, its an excellent model. Very happy with it albeit can see it being quite expensive. For full-stack development its unparalleled.
Secondly regarding the discounted usage - its still showing as the API usage despite what Colin said above.
I raised a thread a few days about the new Billing page but didn’t get a response (I assume its new, if not then apologies).
I assume it was done with good intentions but it is somewhat confusing to me.
I liked the old pricing model - like a fuel guage I didn’t really know the numbers or maths but it was intuitive. Once at 80% and I’ve got a few weeks left, time to start using the Auto model to slow things down a bit.
Now theres 3 numbers, a combined one (how that is calculated I don’t know), an auto one and an API one.
I appreciate I’m fairly small fish, but it causes me alarm when things are changed and not communicated effectively.
I don’t really want to burn through my entire allowance in a week so for now until we get a better understandign or explanation of this page I will continue using the Auto model.
I hope you guys can appreciate my hesitation especially when it comes down to pounds and pence as we like to say in the UK.
2 Likes
Hi Peter!
It can both be true that you’ve used 14% of your API usage, and that Composer is pulling from a different bucket. Check https://cursor.com/dashboard?tab=spending and expand here to see the breakdown.
@Ingvar This is also where should look. There’s no precise amount of Auto + Composer usage, but it’s a generous bucket.
Composer 1.5 (like Composer 1) will bill at API prices against your included API usage, and then on-demand.
1 Like
Our Docs can definitely use some improvement here! It’s on the list.
It’s best to reach out to [email protected] for information specific to your Enterprise, as it can vary by team! Use the e-mail associated with your Enterprise cursor account.
1 Like
I believe this model is likely quite strong—my initial tests have gone very well. It feels a little slower than Composer 1, but that’s probably because many people are trying it out right now.
One point I want to raise is about pricing. It doesn’t make sense to price this model close to Sonnet and higher than Codex. If I could offer advice, it would be to target the price point of grok-code-fast-1, even if that seems unrealistic from a compute and cost perspective at this moment. The goal should not be to price upward, but downward. We need AI to be more affordable to enable large-scale, truly massive deployment.
As it stands, you have to be very selective about what you deploy to see real value. Otherwise, you’re constantly weighing whether to use an agent, handle something manually, or break a task down for a cheaper model—just to keep costs under control. This mindset limits productivity, instead of empowering people to work freely and produce at scale.
Thank you! I got the AI support assistant saying that Enterprise accounts have general pooled usage, so our Composer 1.5 pricing is at API pricing always. Is that accurate?
Hi @Colin
This is how Spending section looks for me.
There’s nothing to expand it seems…
I am not sure if this is true coz I have been mostly using composer 1.5 model today to try it out and my Auto + Composer bucket is still at 1% while my API usage bar is increasing gradually. I tried checking this in the usage in thhe cursor dashboard but that’s not helpful either. It would be better if you could all take some time and improve the billing experience of cursor in general. I can no longer see the dollars consumed in my usage or in the spending either which seems to be weird. Like why are you hiding the dollar usage and resorting to a pure percentage-based thing? I was liking that I was able to see how much of dollars are being spent on API calls as I used my included API usage.
Usage in dashboard should show what bucket in the quota is being used and how much dollars were spent if its API quota at-least, given that every plan has some $X and then some bonus usage.
And not sure if “Composer 1.5 now draws from a combined Auto + Composer request bucket” is true. I am on the new normal pricing plan, not using any legacy request-based pricing mode.
I am so happy to see this, I was waiting on an iteration Composer 1, I decided to share my data with you guys just so you can make a better model on this as It was fast and got most of my tedious tasks done with ease. No it wasn’t the smartest but by far the most usable.
Now on composer 1.5 I’ve been stuck on a problem I had Opus 4.6 and GPT 5.3 Codex on it and both of them failed to understand my repo enough to tackle this issue. But after prompt composer twice Explicit Search and Understand Prompt with an Action Prompt. It handled it first Try.
I am so glad this is actual jump in performance, Its sad to see the cost jump up so high but Its worth it for the time that is saves me. Please don’t keep raising the price I’ve loved Cursor but I was about to Stop my Ultra Sub that I’ve had since Ultra came out. Composer 1.5 is the only thing keeping me here and the fact that Claude Code doesn’t have good indexing and terrible context management.
1 Like
@Colin is that mean user who subscribe yearly on july or before will get unlimited composer 1.5?