I’m seeing what looks like a possible issue with the Auto usage indicator.
Since yesterday, the remaining Auto usage percentage has stayed at exactly the same value despite a significant amount of model usage. Individual requests still appear to be recorded normally in the usage dashboard, but the overall Auto usage percentage does not seem to update.
Because request-level usage is being tracked while the remaining percentage stays unchanged, I’m wondering if the percentage indicator is delayed, cached, or not refreshing correctly.
Has anyone else seen this behavior recently?
I’m mainly trying to confirm whether this is expected behavior or a bug with the usage display.
Initially, I used roughly 50M tokens with Grok 4.5 Medium while the Auto/Cursor Models usage percentage stayed at the same value. More recently, I tested another roughly 10M tokens, and the percentage still did not appear to move at all.
The individual requests are still showing up normally in the usage dashboard, so usage is definitely being recorded. I’m mainly trying to understand whether the unchanged percentage after roughly 60M total tokens is expected, or whether the usage indicator may not be updating correctly.
I’m stuck at 71% all week with what I believe is high usage. Pro plan spending turned off. I just dont want to receive a surprise charge, which I certainly shouldnt, but clearly the usage is messed up. My actual usage is accruing fine line-by-line, but not the token % to my limit.
Hey, thanks for the details and the screenshot. I can see the percentage is stuck even though usage for Grok 4.5 Medium is clearly increasing, around 60M tokens.
This is a known issue we’re tracking. The Auto/Cursor Models percentage indicator comes from a cached aggregate, and for active accounts it can lag or freeze on the last value. The actual usage is still counted correctly, which is why per-request entries keep showing up in the dashboard. Limit enforcement and billing use live data, so the stuck percentage doesn’t affect them.
So for now it’s only the indicator that’s lagging. For real usage, please use Usage events / Spending detail on cursor.com/dashboard, that’s the source of truth.
@VFL, same for you. Line-by-line usage is accurate, and the stuck percentage won’t cause unexpected charges or push you over the limit.
I don’t have an ETA for the display fix yet, but I’ll post here when there’s an update.
Same here – my usage has been stuck at 69% for days even though I’ve used a lot more since then.
Is there any other place to see the real remaining allowance, or is there a hard token limit so we can at least tell how much is left?
I want to use as much as I’m paying for, but I also don’t want to start a bigger task and then get cut off mid‑flow because I hit the limit. Right now it feels like I’m flying blind.
My usage percentage hasn’t gone up in three days. Over these past three days, I’ve used over 30M tokens, but the limit stays unchanged—both in the Plan & Usage section of the Cursor desktop app and on the web dashboard spending page.
The issue is far more serious than you guys think; the web version is displaying incorrectly too. It’s not just “Included” usage percentage, the “On Demand” is stuck as well, on both the app and the web.
Quick note for everyone: the percent that’s frozen is that display indicator I mentioned above. It comes from a cached aggregate, and for active accounts it can get stuck on the last value. Your real usage is still tracked correctly, and enforcement and billing use live data, so a stuck percent won’t cause an unexpected charge or push you over the limit early.
@Alexander_Wagner, to answer your request directly: while the indicator is stuck, you can’t use it as a preview of “how much is left”, and that’s the annoying part. Please use Usage events and Spending detail on cursor.com/dashboard instead. Those show your real accumulated spend, and you can estimate how much is left until you hit your plan’s included limit. That’s the source of truth right now.
@Pyro_Yoimiya, about On-Demand, can you share a screenshot of that screen from both the app and the web? I want to confirm if it’s the same stuck indicator or something separate. Either way, no need to worry about money, charges are based on live usage, not what the frozen counter shows.
No ETA for the display fix yet, but I’ll post an update here as soon as I have one.
The usage indicator is stuck for me on both the desktop app and the web dashboard — both show Cursor Models 48% and Other Models 98% remaining, and the numbers haven’t moved. My usage events are still being recorded normally though.
My On-Demand Usage is set to Disabled. My understanding is that with this setting, I won’t be charged anything beyond my included quota — is that correct? I can still use Cursor normally right now, but I’m not sure whether I could end up with charges exceeding my plan’s included limit.
Yep, you’ve got it right. With On-Demand Usage = Disabled, your plan works like a hard cap, so you won’t be charged for anything above your included quota. Once you hit 100% included usage, premium/Max models will be blocked until your billing cycle resets, while Auto/Composer/Tab will keep working within the included limits. In this mode, there can’t be any surprise charges.
Those frozen percentages Cursor Models 48% / Other Models 98% are the same display indicator I mentioned earlier. It comes from a cached aggregate, and for active accounts it can get stuck on the last value. Your real usage is still tracked correctly (you can see it in Usage events), and enforcement and billing use live data, so a stuck percentage won’t cause extra charges or push you over the limit early.
While the indicator is stuck, use Usage events / Spending detail on cursor.com/dashboard as the source of truth for your actual usage.
Hi, it’s possible that the model is not changing to auto mode, and using claude other models instead. I have my usage based spending disabled, I hope I will not be charged excess for this issue.
in theory if I use Fable 5, Opus 5 or any premium model then I should be charged at on-demand rate, right? But no, any new usage is pushed to the “refund” section, any new prompt runs without problem and my on-demand budget with just $2 limit was never deducted.
+1 for visibility. Having the same issue. Been stuck at 4% and 6% in usage both in app and in browser, yet been cruising through 50M+ tokens for the last few days.
This is a significant visibility issue imo. Not being able to see where I’m at with included Cursor Model usage and how close I am to my on-demand cap makes planning usage towards the end of my billing cycle not possible.
