The Opus 4.5 agent isn’t showing that it consumes 2x requests as other models do. This is misleading. We didn’t realize it used 2x requests until we went and checked our usage. It needs to show the same verbiage that shows on other models that use 2x requests per request.
Steps to Reproduce
Switch to Opus 4.5 model and sent a chat request.
Go and check usage to see 2x requests used for the model.
Expected Behavior
When hovering over the model in the dropdown (and in Models in Cursor Settings), I expect to see the pop out card state that the model uses 2x requests.
Hey, thanks for the report. You’re right - it’s confusing when the UI doesn’t show that the model uses 2x requests.
According to the specs, Opus 4.5 does use 2x fast requests per one request, like other premium models (e.g., Sonnet 4.5). This info should appear in the tooltip when hovering over the model and in Cursor Settings > Models.
I’ll pass this to the team as an important UX improvement - users should see the cost multiplier before using the model, not after checking usage.
Thanks! My team and I are still a bit confused as yesterday Opus only used 1x request for each request, so is there a page where we could get a better understanding of how the requests work (1x vs. 2x requests)? I have some requests with 4M tokens (with 3M cached) that used 1 request, then I have another request with 187k tokens (100k cached) that used 2x requests. Just trying to wrap our heads around how the requests/usage works better.
Thanks for the clarification. Here’s how usage is counted:
How usage is calculated:
Cursor counts usage based on tokens according to the API pricing of models, not fixed “requests.” The 2x multiplier for Opus 4.5 (and other premium models) means the model uses your quota faster - it doesn’t mean each request counts as “2 requests,” but that tokens for this model cost more on the API.
Why requests differ:
A request with 4M tokens (3M cached) and a request with 187k tokens will use different amounts of your quota depending on:
Input tokens volume
Output tokens volume - these cost more
Cached tokens volume - these cost less
The API price of the specific model
Where to see details:
A detailed breakdown of tokens and costs per request is available on your dashboard: Cursor - The AI Code Editor - showing input/output/cache tokens for each request.
You are probably not even a human. Answers are typical AI answers even if you are a human. There is not a place that shows premium models counts as 2x fast request. This must not be a case. It is completely based on api pricing anymore. There is no multiplier anywhere on pricing plans or documentation. There is no such info in models page.
I’m definitely a human. Thanks for the clarification, you’re partially right.
Indeed, the current billing system is based on the API pricing of models, not fixed “requests.”
The term “2x requests” is a simplification we used earlier. The real cost is shown on your dashboard: Cursor - The AI Code Editor - where you can see the token breakdown.
I’ve passed this to the team to clarify model pricing more clearly in the UI.
Thanks for clarifying more on the price and passing the feedback to the team. Hopefully they can expound more on the details of pricing/usage in their documentation rather than having to go into the account/usage every time. Having a “formula” to be able to understand better right off the bat would only help users and prevent feedback/issue requests. Thanks again!
