Plan limits
The limits of the Free, Builder and Agency plans: projects, requests, database and file storage size, email, AI, cron and custom domains.
The numbers on this page come from manh.run’s live plan definitions. For prices and payment, see Pricing.
Account
These limits count across your whole account, all projects together.
| Limit | Free | Builder | Agency |
|---|---|---|---|
| Projects | 3 | 20 | 100 |
Emails sent per month (manh/email) | 200 | 3,000 | 10,000 |
| Emails sent per day | 50 | 500 | 2,000 |
AI tokens per month (manh/openai) | 500,000 | 3,000,000 | 12,000,000 |
Per app
| Limit | Free | Builder | Agency |
|---|---|---|---|
| API requests per month | 100,000 | 1,000,000 | 5,000,000 |
| Database size | 100 MB | 1 GB | 5 GB |
| File storage size | 1 GB | 10 GB | 50 GB |
| Cron jobs | 5 | 20 | 20 |
| Custom domain | No | Yes | Yes |
| ”Made with manh.run” chip on the app | Yes | No | No |
Months are calendar months in UTC: the request count resets on the 1st of each month.
Per request
| Limit | Free | Builder | Agency |
|---|---|---|---|
| CPU time per request | 5 s | 10 s | 30 s |
| Outgoing calls per request | 100 | 500 | 1,000 |
| Requests per minute | 600 | 3,000 | 10,000 |
| Live connections at once | 100 | 1,000 | 5,000 |
When you reach a limit
- A running app doesn’t stop. Only the part over its limit answers with a clear error. For example, when emails run out, sending an email fails while the other APIs keep working.
- Monthly requests and database size never slow the app down or block it. At 80 % and at 100 % you get one email each month; if an app stays far above its plan, manh.run gets in touch.
- A paid plan that expires doesn’t charge you automatically. You have 7 days to renew; after that, the account moves to Free.
- On Free, your apps keep running and no data is deleted. You can’t create projects above the limit, custom domains are paused and apps show the manh.run chip.
- After a downgrade, only the newest projects within the new plan’s limit can be deployed.
Good to know
- Outgoing calls are the
fetchcalls your server code sends during one request. A loop that calls an outside API once per row hits this limit first. - See what you’ve used under Settings › Usage in your project.
- Email and AI count across your account: one app that sends many emails leaves fewer for your other apps.
- Email counts recipients; a day runs on UTC time.
- AI counts tokens, not calls: an input token of the large model counts 1, an output token counts 2; the small model counts half; embeddings count 1/20.
Next steps
- Pricing — the price of each plan and how to upgrade.
- Data and security — how data and rollback work.
- Scheduled jobs — write a
.cron.tsfile.
Last updated