VM1
Sold out
$ 13,455/ mo
- 1 vCPU
- 1 GB RAM
- 15 GB SSD
- 1 static Argentine IP
- Unmetered transfer (fair use)
- Windows or Linux
- AI supervision around the clock
Codex VPS
OpenAI’s Codex CLI on a Linux VPS with root: leave it a job in tmux or scheduled with cron, and check it when you come back. Code and credentials stay on the server, in Buenos Aires.
In short
OpenAI’s Codex CLI installs on a Linux VPS with one command and signs in with the customer’s account, even with no desktop, using `codex login --device-auth`. It runs interactively or unattended with `codex exec`, so it suits scheduled jobs. OpenAI asks for 4 GB of RAM and recommends 8: the VM4 ($ 53,820 a month) meets the minimum and the SH2 cloud server gets close to the recommendation.
The ChatGPT account or OpenAI API key is the customer’s and stays in their name. We provide the server, the IP and the operating system. The server is delivered unmanaged: we give you the commands from the official documentation and a VPS ready to run them, but installing, configuring, updating, backing up and securing the application is up to you or your technical team. We help you get started and with specific questions, and that does not replace administering the server.
Codex VPS
With codex exec the agent runs with nobody watching the screen: schedule it with cron or leave it as a systemd service.
In tmux the job keeps going even if the connection drops, and you pick it up from any machine with one command.
The repository, the credentials and the generated files live on the VPS, in our Buenos Aires datacenter, not on a laptop that gets lost or broken.
If your application is on our infrastructure, the agent works on the same network, without taking data outside.
Plans
Final monthly price, no extra taxes or hidden fees. USD amounts are the official store list; local invoicing is issued in pesos.
Sold out
$ 13,455/ mo
$ 26,910/ mo
$ 40,365/ mo
Most chosen
$ 53,820/ mo
Getting started
Log in as root, update and add git and tmux.
apt update && apt install -y curl git tmuxOpenAI’s official installer leaves the binary ready to use.
curl -fsSL https://chatgpt.com/codex/install.sh | shDevice authentication shows a code that you confirm from your phone or computer. Then tmux new -A -s codex for the persistent session.
codex login --device-authCPU24%
23,7% / 4 vCPU
MEMORY16%
649 MB / 4 GB
DISK39%
23 GB / 58,8 GB
How to connect
ssh root@203.0.113.24
Example view with fictitious data.
Control panel
Power on, shut down, restart or reinstall the operating system from your browser, and if the server network does not respond, get in through the rescue console. CPU, memory, disk and network usage stay in view by hour, day, week or month.
Questions
One of the two, and you bring it: Codex signs in with a paid ChatGPT account or an OpenAI API key. The credential is stored on the VPS, in your name.
With codex login --device-auth: the terminal shows a code that you confirm from the browser on your phone or computer, with nothing to open on the server.
Yes. codex exec runs without interaction, so you can call it from cron or a systemd service and leave the result in a file or a ticket.
OpenAI asks for 4 GB of RAM and recommends 8. The VM4 ($ 53,820 a month) meets the minimum; if you are going to build or run containers alongside, an SH2 cloud server or above is a better fit.
The VPS is delivered with a clean Ubuntu 24.04 and the three commands on this page get it running. The server is unmanaged: installation, updates, backups and scheduled jobs are up to you or your team, as clause 22 of the terms and conditions states. If you get stuck, write to us and we will point you in the right direction.
Keep exploring