Cursor Says You're Out of Usage
First figure out which pool you hit
As of 2026-10-07, Cursor splits usage into two separate pools, Cursor Models (Grok 4.7, Grok 4.6, Grok 4.5, Composer 2.5) and Other Models (third-party models at API price), each resetting with your monthly billing cycle. Confusing which pool you hit, plus assuming the reset date is the 1st of the month, are the two most common mistakes.
Updated 2026-10-08
Hit your subscription limit or keep getting 429s? Switch to an API billed per token and keep working with one key.
- Billed per token — live rates on /models
- Pay by card (Visa / Mastercard / AMEX), Apple Pay, Google Pay or crypto
- Self-service: top up and activate instantly
Four key facts
Two separate usage pools
Cursor Models (Cursor's own optimized models) and Other Models (third-party, billed at API price) are metered separately and don't offset each other.
In the Cursor Models pool, not Other Models
As of 2026-10-07, Cursor's official docs list Grok 4.7, Grok 4.6, Grok 4.5 and Composer 2.5 in the Cursor Models pool, which has significantly more included usage — the common claim that 'Grok 4.6 eats the Other Models pool' isn't accurate.
Reset date = your subscription date, not the 1st
Individual plans reset on your subscription anchor (subscribe Jan 15 at noon, and it resets on the 15th each month at noon); team plans reset on a shared team billing date instead.
Other Models supports pay-as-you-go
Pro / Pro Plus / Ultra all include an Other Models pool; once it's spent you can add on-demand usage at the same API rates or upgrade. As of 2026-10-07 the India-only Start plan (₹649/mo) has no Other Models pool.
What the error text is actually telling you
The full message is usually "You've reached your usage limit" — it doesn't say which pool. As of 2026-10-07 the Cursor Models pool covers Grok 4.7, Grok 4.6, Grok 4.5 and Composer 2.5 (plus their fast variants) with significantly more included usage. The Other Models pool covers every third-party model, including Claude, Gemini, and GPT families, billed at each model's API price. The two pools are metered independently — running out of Other Models doesn't touch your Cursor Models balance, and vice versa.
Latest development
Cursor has raised included-usage limits on individual and team plans multiple times through 2026, but the two-pool split and subscription-date reset haven't changed. As of 2026-10-07 the official docs add Grok 4.7 to the Cursor Models pool and list Start, an India-only plan (₹649/mo) that covers the Cursor Models pool but not Other Models. To tell which pool you hit, check the Spending tab in your dashboard — it shows each pool's consumption and reset time separately.
Timeline
The Cursor Models / Other Models split and subscription-date reset have been in place since the plan system was built.
Cursor raises included-usage limits for individual and team plans several times; the two-pool split and reset rules stay unchanged.
Checked against the official docs: the Cursor Models pool now lists Grok 4.7 alongside Grok 4.6, Grok 4.5 and Composer 2.5; the 'Grok 4.6 eats Other Models' misconception still circulates online.
Confirmed vs. common misreading
Confirmed
Cursor's official docs spell out both pools and which models belong to each (as of 2026-10-07 Grok 4.7 / Grok 4.6 / Grok 4.5 / Composer 2.5 sit in the Cursor Models pool; a selected third-party model draws from Other Models at that model's API rate); individual plans reset on the subscription anchor, team plans on a shared team billing date.
Common misreading
A claim circulating online says Grok 4.6 eats the Other Models pool — that contradicts the official docs. Cursor Grok 4.6 sits in the more generous Cursor Models pool and doesn't touch third-party model quota.
How the two pools differ
Cursor Models pool
As of 2026-10-07 covers Grok 4.7 / Grok 4.6 / Grok 4.5 / Composer 2.5 (and their fast variants); included in your plan, usually generous. The India-only Start plan includes only this pool.
Other Models pool
Covers Claude, Gemini, GPT and every other third-party model, billed at API price; once spent you need on-demand billing or a plan upgrade to keep going.
How to confirm which pool you hit
Open the Spending tab in your dashboard — both pools' consumption, remaining balance, and reset time are shown separately; as of 2026-10-07 the official docs also say both pools are visible in your editor settings and on the usage dashboard. Check which model you were using when the error happened: Grok 4.7/4.6/4.5/Composer 2.5 usually means the Cursor Models pool is full; anything else usually means Other Models. The reset time is your subscription anchor date and time, not the 1st of the month (except on team plans).
What to do on QCode
If it's the Other Models pool that's spent, one QCode key calls the Claude, GPT and Chinese model APIs directly at official price × service rate, independent of Cursor's plan pools and without waiting for Cursor's reset date.
FAQ
Which pool does Grok 4.6 actually draw from?
The Cursor Models pool. As of 2026-10-07 Cursor's docs list Grok 4.7, Grok 4.6, Grok 4.5 and Composer 2.5 there (including fast variants), included in your plan and fairly generous. It would only fall into Other Models if you called a non-Cursor-optimized Grok build through some other channel.
Why isn't my reset date the 1st of the month?
Individual plans reset based on the date and time you subscribed, not the calendar month. That means your usage cycle and finance calendar can drift out of sync — that's the intended design, not a bug.
Can the two pools borrow from each other?
No. They're metered independently — exhausting Other Models doesn't touch your Cursor Models balance, and vice versa.
Do team plans reset the same way?
No. Team plans reset for all members at once based on the team's billing cycle, not each person's individual subscription date.
What can I do besides waiting for the reset when Other Models runs out?
Enable on-demand pay-as-you-go billing in the dashboard, upgrade to a higher plan for more included Other Models usage, or route that work through a different entry point.
What's the fastest way to tell which pool caused this error?
The model you were using when it happened is the clearest signal: Grok 4.7/4.6/4.5/Composer 2.5 → Cursor Models pool; anything else → Other Models pool (per Cursor's docs as of 2026-10-07). Then confirm the balance on the Spending tab.
Sources
The two-pool split, which models belong to each, and the subscription-date (not calendar) reset anchor come from Cursor's official documentation (Models & Pricing, Usage and limits), first checked on 2026-08-27. The Grok 4.7 assignment, the Start plan and the monthly-billing-cycle reset wording were re-checked against Models & Pricing on 2026-10-07.
Don't wait for the reset when a pool runs dry
One QCode key calls Claude, GPT, Chinese models and more directly — independent of Cursor's plan pools.
Related reading
Claude Weekly Limit 2026: the full rules
A different quota mechanism explained in full, useful for comparison.
How to rotate quota across multiple tools
Rotating usage across Claude Code / Codex / Cursor sensibly.
Cline 400 with a custom Anthropic base URL: the fix
4.1.22 started sending the fallbacks option and custom endpoints returned 400; 4.1.23 (2026-10-07) fixes it. Plus the QCode Anthropic-provider setup.
This page is technical explanation, not an official Cursor statement. Specific quota rules and reset times are governed by your Cursor account dashboard and official documentation.