I’d like to know the usage limits for the latest Claude models available with the Pro membership—is it a small amount, a generous amount, or the same as the official Claude subscription?
Hey, thanks for the question. On Pro, there isn’t a direct comparison to the official Claude subscription since the pricing model is different.
In short: Pro 20 USD/month includes about 20 USD of usage per month, and it’s not spent by number of requests. It’s spent based on the API token cost of the specific model. So how long it lasts depends on which Claude model you’re using:
- Light use (Tab, occasional Agent) usually fits within the included 20 USD.
- Daily Agent users usually end up around 60 to 100 USD/month.
- Power users (multiple agents, automation) are 200 USD/month or more.
Opus is the most expensive Claude model, so the included usage runs out noticeably faster than with Sonnet or Composer. Once the included usage is used up, you can either stop or enable on-demand usage-based pricing and keep paying at the API rate.
Model and pricing details: Models & Pricing | Cursor Docs
How usage vs included is calculated: Models & Pricing | Cursor Docs
Let me know if you want me to estimate it for your specific usage scenario.
The useful way to think about it is workload, not request count.
If Claude is mostly for review, planning, or occasional hard debugging, the included usage can last a while. If it is the default agent model for daily coding, especially with long contexts, it will behave much more like API spend than a fixed subscription bucket.
Good add, @Jason_Luo1, that’s exactly the right way to think about it. How long it lasts depends less on the number of requests and more on which model you use and how much context you’re sending. Review and planning burn included usage slowly, but the default agent on long contexts behaves closer to API-style spend.
@jsl-7, if you want, describe your typical setup (which model, how often you use Agent, rough task size), and I can estimate how far the included $20 will go.