The question was posted but has been hidden.
@condor plz check
Tokens on the pro plans do not equate to a quota, one can not think of them as a metric to gauge usage. There’s a few factors like amount of requests, the type of requests, services involvement, and models that all come into play. This is also why you can have 10million token usage 1 month and the next have 20million.
You should only think of tokens as amount of text processing which is why it’s a good idea to review your request’s token usage and context. These give a hint of “are my requests pulling in too much context”. They are in a sense a metric of efficiency.
Instead of just letting your agent review the entire file to solve a problem in a small block of code, do a CTRL+L on the small block of code into cursor agent window and talk about it. The agent will then see if it can answer with the small amount you provided. If you pull in huge files and reference them all, the agent will process EVERYTHING as you requested so token usage will go much higher…even if the actual work is very small.
I’ll leave you with a tip, use the new chat window more often. Unless you really need to continue a conversation on something, new chat window is better.
In my testing scenarios, even within the same conversation in the same folder, the Pro plan under the student package consumes its periodic quota at three times the rate of the regular plan.
I’m not a new user. I understand how context and caching affect consumption. But I have sufficient evidence to show that the student Pro plan has been systematically reduced, rather than the difference being caused by usage patterns.
The official side knows this, which is precisely why they don’t dare to give me a direct answer.
well, it can get tricky because of promotion, when did you tested composer 2.5 fast on your pro account? like in the past 2 weeks, there have been at least, at least 4 period of time in which composer had massive amount of discount (i’m not talking only of the discount for using the sdk), also, you can actually check the details of cache read, input, output, … by hovering over to see the details, also man :(( like, if your goal is to maximize usage, you should stick to the non fast variant, it’s literally 6 time cheaper for only half the speed
dude I processed the same issue using the same empty session on the same day with roughly the same number of tokens and the same model. Normally, Pro shows 1% usage, but Student Pro shows 3%. How do you explain this?
Please stop changing the subject. The current issue is that, under reproducible conditions with controlled variables, the Student Pro plan only has one-third of the normal Pro Auto + Composer usage. Please focus on resolving this issue. I can give you my student account so you can test it yourself.
Let me be clear about what this is and isn’t. I don’t care whether students use more or less than anyone else — that’s not the issue. I pay full price for Cursor myself, so this isn’t some student trying to squeeze out more free stuff.
The issue is the marketing. Cursor sells the student plan as Pro — same plan, “no difference.” But the Auto + Composer allowance is quietly capped at a third of what paid Pro gets. That’s not “a free year of Pro.” That’s a worse plan wearing Pro’s name.
If they want to give students a reduced tier, fine — say so. Put “Auto + Composer limited to ~1/3 of Pro” right next to “free year of Pro,” and let people decide with the real numbers in front of them. What they shouldn’t do is advertise one thing, ship a third of it, and count on nobody comparing two accounts side by side.
of course man, I’m not trying to change the subject, 我们会一起渡过难关的，我会尽力去打听相关情况。很抱歉造成不便，我刚才只是想彻底弄明白整件事。
I’m not coming at you, bro. I really appreciate you paying attention to this issue and trying to figure it out.
yay! well, i’m searching on my own right now, but I will also get the team notified, and will get back with the answers!
also @Dbxnblg 
I want to reassure you, the reason your past thread was hidden is only because it was flag (automatically) for a billing issue and sent off to [email protected] (it’s them who usually handle that), I understand why you felt gaslighted by the system, it happened to me to
any updates bro？
Hey there.
Student Pro and Pro share the exact same plan limits. There isn’t a hidden student cap.
If you can grab two request IDs, one from each account, with privacy mode disabled, I’m happy to look into exactly why they contributed to your quota differently!
Let’s keep things good-faith here. I know it’s frustrating when the numbers don’t seem to add up, but we’re not hiding anything or dodging questions.
I ran the two accounts as a controlled test, specifically to rule out usage habits or model differences. Same model (Composer 2.5-fast), same prompt, same mode (chat), near-identical token counts. Here’s what came back:
Student account — request 7557131b-b19c-4ad6-9592-42208919095c: 250k tokens, 0.4% of allowance.
Paid Pro account — request 5a862c5d-19ab-4783-9e86-a2d9e7467199: 230k tokens, 0.1% of allowance.
Same job. The student account actually burned 4x the share of my allowance. The only variable between these two requests is which account ran them.
You can pull both request IDs and verify it yourselves. And to be clear, I’m not assuming bad faith: if this 4x gap wasn’t set on purpose, then there has to be a reasonable explanation for it. So what is it?
Based on the information I’ve got, it’s not just me. As long as you have both a paid Pro subscription and a Student Pro subscription at the same time, you’ll notice this problem.
@Tom_Coustols @Colin been 3 days. wonder how’s it going
@kevinn I hope this issue will receive official attention. More and more students have reported encountering the same problem as I did.