Skip to content

Account

Limits & fair use

Per-plan limits, rate limits, network rules (port 25 is blocked) and what counts as abuse.

Limits keep the platform fast and fair for everyone and cap the damage if an account or API key is misused. If you need more than your plan allows, tell us.

Plan limits #

CPU, memory and disk are the sizes of all your team's services added up.

Hobby Pro Enterprise
Projects 3 10 50
Services 10 50 200
Services per project 10 25 50
Service sizes Small, Medium, Large Small, Medium, Large, XL Small, Medium, Large, XL, 2XL
CPU, total 4 vCPU 16 vCPU 64 vCPU
Memory, total 8 GiB 32 GiB 128 GiB
Container disk, total 32 GiB 128 GiB 512 GiB
Volumes 10 25 50
Largest volume 10 GiB 50 GiB 100 GiB
Volume storage, total 10 GiB 50 GiB 100 GiB
Live log streams at once 10 20 50
Consoles at once 2 3 5

API rate limits #

The REST API and the MCP server share these limits:

  • 60 requests per minute per API key, and 300 per minute per team.
  • 10 deploys and redeploys per minute per key.
  • 30 other changes (create, update, start, stop, delete) per minute per key.
  • 30 image checks per minute per team (each create or deploy with a new image checks the registry).

Over a limit, you get HTTP 429 with a Retry-After header.

Network rules #

  • Services can connect to the internet, with these exceptions.
  • Outbound port 25 (SMTP) is blocked. Send email through an email provider (Postmark, Mailgun, Amazon SES, …) on its submission port, 587 or 465.
  • Private and internal network ranges are not reachable, and known abusive destinations are blocked.
  • Outbound traffic from each container is limited to about 50 Mbit/s sustained; short bursts can go faster.
  • Services in different projects can't reach each other; use one project for services that talk to each other.

Fair use and abuse #

You can run almost anything legal. Not allowed, among others (the full list is in section 5 of the Terms):

  • Cryptocurrency mining and other workloads whose purpose is to burn CPU;
  • spam, open mail relays and open proxies;
  • phishing, malware, botnets and command-and-control servers;
  • scanning or attacking other systems, or trying to escape the sandbox.

We watch for patterns such as a service using its full guaranteed CPU or more for hours, or unusually large outbound traffic. A flag is reviewed by a person before anything happens; if a service clearly breaks the rules, we may stop it and tell the team owner why.

To report a service hosted on QuickTane, use the abuse form or write to [email protected].

Something missing or wrong? Tell us — or open a ticket from the dashboard.