Your use of Moon AI is measured over three windows: a 5-hour session, the week and the month. Each one is shown as a percentage, with the exact time it goes back to zero. What each window includes depends on the workspace’s plan: see the plans.
The three windowsLink to the The three windows section
| Window | When it starts | When it resets |
|---|---|---|
| “Session” | With your first message after the previous session ends | Fully, 5 hours after it started |
| “Week” | With your first use after the previous week ends | Fully, 7 days after it started |
| “Month” | At the start of the current period; on a paid plan, your billing period | At the end of that period |
The windows don’t slide: a session that starts at 11:40 AM resets all at once at 4:40 PM, no matter when you used Moon AI in between. On a paid plan, the monthly limit follows your billing period, not the calendar month.
Everything you do counts toward these windows: messages and their answers, images, videos, voice, web search and agent actions. As soon as one of the three is full, new requests wait until it resets.
See where you standLink to the See where you stand section
| Where | What it shows |
|---|---|
| The small moon on your account button, at the bottom of the sidebar | The percentage of the fullest window; on hover, “Usage limit · …% used · resets …” |
| Account menu › “Subscription and usage” | Under the entry, the percentage of the fullest window, and on a computer when it resets (“Session · resets at 4:40 PM”, for example) |
| The “Subscription” page, “Usage limits” card | All three windows: “Current session”, “Week” and “Month”, each with its percentage and reset time; “Refresh usage” reloads the numbers |
| “Your month” tab of the “Subscription” page | The share of the monthly limit used and its renewal date, with usage by model |
| “Usage” tab of the “Subscription” page | Your detailed usage in tokens, by model and by use, with the share of the monthly limit for each |
The “Subscription” page opens from the account menu › “Subscription and usage”, from “Settings” › “Subscription and usage” or at chat.realmoon.ai/billing; in a managed organization, the entry is called “License and usage”. It reminds you that “A session starts with your first message and lasts 5 hours.”
Read the reset timeLink to the Read the reset time section
| Display | Meaning |
|---|---|
| “in 38 min” | In less than an hour |
| “at 4:40 PM” | Today, at that time |
| “tomorrow at 9:05 AM” | Tomorrow |
| “Thursday at 10:00 AM” | In the next few days |
| “on November 1” | For the month, the renewal date |
| “Starts with your next message” | No session is running |
When you reach a limitLink to the When you reach a limit section
A refused request is never lost: Moon AI writes a message in place of the answer, with the window concerned and when it comes back.
| Full window | Message |
|---|---|
| Session | “You have reached your session limit.” |
| Week | “You have reached your weekly limit.” |
| Month | “You have reached your monthly limit.” |
| Request larger than the plan allows | “This request is larger than your plan allows.” |
Each message is followed by “Your limit resets…”, with the time. The “Try again” button (or “Resume” if some actions were already done, without running them again) comes back at that time. If the limit is reached during a long answer, the answer stops with “Usage limit reached. The answer stops here.” and “Continue” comes back with the limit.
The home screen shows it too, with a card such as “Your session is taking a break.”, when it renews, and “View my limits”. Once the limit is back: “Your limit has reset.”
From 90% of a window, the “Subscription” page warns you: “You have used…% of your… limit”.
More roomLink to the More room section
Depending on your situation, the message offers:
- to the workspace owner, “View plans” or “Upgrade to…”, when a higher plan exists;
- to a member of a team workspace: “Ask the workspace owner to change the plan.”;
- in a managed organization: “Ask your organization’s administrator if you need more.”
For the monthly limit, if you can change plans, the plans window can also open on its own, at most once a month, unless you have paused plan offers: see Plan offers. You can compare plans on the plans page; you change plans from the “Subscription” page: see Subscription and billing.
What fills your windows fastestLink to the What fills your windows fastest section
The cost of a request depends mostly on the model and on how much text it reads and writes.
| What uses more | Why |
|---|---|
| “High cost” and “Very high cost” models | Each message costs more to process |
| Long conversations and large documents | The model rereads everything that came before with each message |
| Agent mode | One request chains several steps and tools |
| Images and videos | Creating them costs more than text |
| Several models mentioned with @ | Each answer counts toward your limit |
Deep research doesn’t count toward the 5-hour session, only toward the week and the month.
To make your limit last: pick a “Low cost” model for simple questions, start a new conversation when you change topics, and keep “High cost” models for tasks that need them.
Very high cost modelsLink to the Very high cost models section
In the model picker, “All models” shows three cost bars: “Low cost”, “Medium cost” or “High cost”. A “Very high cost” model (red bars) uses a large share of your 5-hour limit with each message; the percentage appears on hover, for example “One message uses about…% of your 5-hour limit”.
When you pick such a model, a notice appears above the message box: “… uses a lot, about…% of your 5-hour limit per message.” In agent mode, it adds that one request often uses several times more.
- “Switch to Auto” (or “Let Auto choose”) goes back to automatic model selection.
- “Hide for this model” hides the notice.
The notice is for information only: you can still send your message. See also Choose a model.