*[DediGPU](https://dedigpu.com/) — Markdown mirror of [https://dedigpu.com/docs/limits-and-fair-use](https://dedigpu.com/docs/limits-and-fair-use) · updated 2026-09-02 · index for LLMs: [llms.txt](https://dedigpu.com/llms.txt) · everything: [llms-full.txt](https://dedigpu.com/llms-full.txt)*

# Limits and fair use

> The few limits on a DediGPU server: no throttling of GPU load, unmetered ports, port 25 closed by default, eight extra IPv4, five SSH keys, the acceptable-use policy in short.

It is your server; we do not throttle it. The limits below exist because of the network, the hardware or the law.

- **GPU load:** No limit. The cost sheet assumes a 70 % fleet average; a server at 100 % all month is fine.
- **Network:** Unmetered at port speed. Outbound port 25 closed by default; opened on request with a purpose.
- **Addresses:** One IPv4 included, up to eight more per server. IPv6 /64 on request.
- **SSH keys:** Up to 5 per server at order time; any number in the key ring.
- **Servers:** Up to eight cards per server; 16 to 512 cards as a [cluster](https://dedigpu.com/clusters).
- **Top-ups:** $40 minimum, $20,000 per invoice, several invoices at once.
- **API:** 60 requests per minute per token.
- **Support:** Open to accounts with a server online. Same support for everyone.

## Acceptable use, in short

Training, serving, rendering, mining and full-power workloads are what the servers are for. Attacks, malware, phishing, spam, sexual content involving minors, content illegal where the server sits and intimate imagery of real people without consent are not allowed. Reports are forwarded to you with 24 hours to answer; ongoing harm is stopped by suspension. The full text: [acceptable-use policy](https://dedigpu.com/legal/aup).