Usage based pricing incorrect and not well documented

Describe the Bug

The usage based pricing column says “Requests” in the header, but shows $ in the units. Either way, the numbers do not make sense, and is also not consistent with that it was before.

Note that MAX MODE is not enabled.

Steps to Reproduce

  1. Enable usage based pricing
  2. Use the models as expected

Expected Behavior

  1. The requests should cost $0.04 per per request
  2. The column should show the price for each request

Operating System

MacOS

Current Cursor Version (Menu → About Cursor → Copy)

Version: 1.1.6 (Universal)
VSCode Version: 1.96.2
Commit: 5b19bac7a947f54e4caa3eb7e4c5fbf832389850

Does this stop you from using Cursor

No - Cursor works, but with this issue

Hi @conradkoh

thank you for the detailed bug report.

Yes the column should be renamed for current plans.

Docs got updated in meantime, with a section on usage based pricing.

Usage based pricing is calculated based on the underlying compute used for the request and is charged based on model API prices.

You can mouse over every charge and see the breakdown in tokens in/out/cache write/cache read. As the charge is not per request but per API usage the request price isnt $0.04. however the request price is shown as per your screenshot.

hi @T1000 thanks for the response.

I’m sorry but I don’t think it’s right that this is billed in this way unless it is properly communicated. Here are some links to problems with the docs.


link to source - Cursor – Models & Pricing


link to source - Cursor – Plans & Usage

Thats for legacy plans that apply to Team and Enterprise accounts or Pro accounts that switched to legacy Pro plan.

But yes I will forward the feedback to Cursor team.

The single/multi invoice descriptions are independent of the plan and have nothing to do with the request cost.

Thanks for clarifying. The general feedback is that there’s a gap for those who were used to usage based pricing and now the same term means different things in different contexts.

1 Like

Hi T1000

Does model usage based prices same as published rate from each provider or cheaper?

The pricing is from what Cursor reported the API pricing + 20% for cost incurring within Cursor infrastructur.

The closest pricing I can point you to is listed in Docs, though as docs are being improved use following steps until they are adjusted.

  • Cursor – Models & Pricing > Max Mode
  • Scroll down to Overview > Switch to Teams Pricing, then Enable Max Mode to see cost in requests per MTok
  • Take cost in requests / Mtok , divide by 1000 000 and multiply with tokens used (Dashboard > Usage) then multiply by $0.04 (from old calculation) to get to current cost per request logged.

So overlal it would be

  • Input tokens * Input req/1MTok / 1000 000 * $0.04
  • Output tokens * Output req/1MTok / 1000 000 * $0.04
  • Cached Input tokens * Cached Input req/1MTok / 1000 000 * $0.04
  • Cache Write tokens * Cache Write cost /1Mtok (from API providers list + 20%) / 1000 000
    Those added up are total price.

[Edit: the docs are being updated, please a bit patience, it will become simpler to understand usage based pricing]

1 Like

I don’t think it’s fair to ask for patience when this is the current billing structure. Updating pricing clearly is not a good to have feature.

(I don’t mean to be abrasive - but I think these are unmet basic expectations.)

1 Like

@Yurist_Mospravczentr That depends on what you do, what you model settings are in Chat. It can be correct. There is no issue listed unless you explain it.

@conradkoh I would rather that things get improved than left unchanged and users asking the same questions all over :slight_smile: . Not sure if I understand you right, you do not want pricing to be clearer in the docs? hehe

I meant that it is not a “feature” - it is a “bug”. It’s like going to the restaurant, getting billed and they just tell you the total bill but say “please, be patient, we’re working on our docs. pay me now, and come back next time to see the price”.

Not the best analogy, but you can see why this is problematic. No one bills for something if they haven’t made the pricing available for people to see.

The patience should be been exercised by the team before releasing, not an expectation placed on me, seeing that I’m on the receiving end of the bill.

To a large degree - wondering if this is even legal :thinking: Anyways, not that big a deal for me personally, but the team really needs to figure this out. Thanks for responding quickly tho, appreciate it :+1:t3:

2 Likes

@Yurist_Mospravczentr thank you for your posts here. Yes Claude 4 Sonnet is great.

Your screenshot is normal, it shows usage included in the plan as ‘-’ (you are not charged additional cost).

Could you create a separate post about your own usage as you may need more information how the plans work and this thread is not about usage within a plan.

@Yurist_Mospravczentr Oh wow, is it really that bad? I got in a lot more usage than what you’re saying before I got rate limited.

If you go to Cursor - The AI Code Editor, do you see this option “Opt Out of New Pricing”?

Wanted to bring to attention this reddit post as well. Sounds like people are affected way worse than I was.

… and they are shadowbanning complains xD ahahah
resigned, went to windsurf