AI coding VPS

An AI coding VPS is a virtual private server with the Claude Code, Codex, and Grok coding agents preinstalled. Connect over SSH from your terminal or your IDE, and let the agent read your project, run commands and tests, and edit files. The work runs on a server separate from your personal computer.
  • Prices include VAT
  • From $11.43/month
  • Claude Code, Codex, and Grok preinstalled
  • Full SSH access

AI coding VPS plans

STARTER

Suitable for small projects, such as scripts and bots, and for getting to know coding agents

$11.43/month

  • CPU cores2
  • Dedicated RAM2GB ECC
  • Disk space50GB
  • TrafficUnlimited
  • IP address1
  • Claude Code, Codex, and Grok: Preinstalled
  • Snapshot backup: One per month
  • BASIC management: Included
  • Setup fee: Free
  • Server location: Israel
Order STARTER

BASIC

Recommended

Suitable for most developers, for day-to-day work with a coding agent on an active project

$28.57/month

  • CPU cores2
  • Dedicated RAM4GB ECC
  • Disk space100GB
  • TrafficUnlimited
  • IP address1
  • Claude Code, Codex, and Grok: Preinstalled
  • Snapshot backup: One per month
  • BASIC management: Included
  • Setup fee: Free
  • Server location: Israel
Order BASIC

PRO

Suitable for several active projects, databases, and running tests alongside the agent

$57.14/month

  • CPU cores3
  • Dedicated RAM8GB ECC
  • Disk space200GB
  • TrafficUnlimited
  • IP address1
  • Claude Code, Codex, and Grok: Preinstalled
  • Snapshot backup: One per month
  • BASIC management: Included
  • Setup fee: Free
  • Server location: Israel
Order PRO

BUSINESS

Suitable for several agents running in parallel, or a small team working on the same server

$114.29/month

  • CPU cores4
  • Dedicated RAM12GB ECC
  • Disk space250GB
  • TrafficUnlimited
  • IP address1
  • Claude Code, Codex, and Grok: Preinstalled
  • Snapshot backup: One per month
  • BASIC management: Included
  • Setup fee: Free
  • Server location: Israel
Order BUSINESS

ENTERPRISE

Suitable for heavy build and test processes that use many cores in parallel

$142.86/month

  • CPU cores8
  • Dedicated RAM16GB ECC
  • Disk space300GB
  • TrafficUnlimited
  • IP address1
  • Claude Code, Codex, and Grok: Preinstalled
  • Snapshot backup: One per month
  • BASIC management: Included
  • Setup fee: Free
  • Server location: Israel
Order ENTERPRISE

PREMIUM

Suitable for memory-intensive environments, such as large code repositories and databases

$257.14/month

  • CPU cores8
  • Dedicated RAM32GB ECC
  • Disk space400GB
  • TrafficUnlimited
  • IP address1
  • Claude Code, Codex, and Grok: Preinstalled
  • Snapshot backup: One per month
  • BASIC management: Included
  • Setup fee: Free
  • Server location: Israel
Order PREMIUM

Included in every plan

  • Long tasks - keep running even when your computer is off
  • Personal support via HelpDesk
  • BASIC management - included
  • Full root access over SSH
  • Traffic - unlimited
  • Snapshot backup - one per month
  • Firewall - preconfigured
  • Migration from another provider - free
  • AI service connections - supported

Tip from our team

As a leader in cloud services, we have learned over the years that it is always best to start with a small plan rather than ordering the largest plan right away, even for a large business.

Our VPS plans are built to grow with your business: at any stage, you can add CPU cores, memory, and disk space as your needs change. Every plan comes with no commitment beyond the current month of service, no fine print and no cancellation fees.

Not sure which plan to choose? Talk to us, and we will build a plan tailored to your business needs.

Full technical specifications

The specifications are grouped into nine topics: infrastructure, resources, network, operating system, coding agents, security, backups, monitoring, and support. Choose a topic to see all its components.

Included in every plan

  • Long tasks - keep running even when your computer is off
  • Personal support via HelpDesk
  • BASIC management - included
  • Full root access over SSH
  • Traffic - unlimited
  • Snapshot backup - one per month
  • Firewall - preconfigured
  • Migration from another provider - free
  • AI service connections - supported

Plan comparison

AI coding VPS

STARTER

  • CPU cores2
  • Dedicated RAM2GB ECC
  • Disk space50GB
  • TrafficUnlimited
  • IP address1
  • Claude Code, Codex, and GrokPreinstalled
  • Snapshot backupOne per month
  • BASIC managementIncluded
  • Setup feeFree
  • Server locationIsrael
  • Operating systemA range of Linux distributions, including AlmaLinux, Rocky Linux, Ubuntu, and Debian
  • FirewallIncluded
  • Graphs and monitoringIncluded

BASIC

  • CPU cores2
  • Dedicated RAM4GB ECC
  • Disk space100GB
  • TrafficUnlimited
  • IP address1
  • Claude Code, Codex, and GrokPreinstalled
  • Snapshot backupOne per month
  • BASIC managementIncluded
  • Setup feeFree
  • Server locationIsrael
  • Operating systemA range of Linux distributions, including AlmaLinux, Rocky Linux, Ubuntu, and Debian
  • FirewallIncluded
  • Graphs and monitoringIncluded

PRO

  • CPU cores3
  • Dedicated RAM8GB ECC
  • Disk space200GB
  • TrafficUnlimited
  • IP address1
  • Claude Code, Codex, and GrokPreinstalled
  • Snapshot backupOne per month
  • BASIC managementIncluded
  • Setup feeFree
  • Server locationIsrael
  • Operating systemA range of Linux distributions, including AlmaLinux, Rocky Linux, Ubuntu, and Debian
  • FirewallIncluded
  • Graphs and monitoringIncluded

BUSINESS

  • CPU cores4
  • Dedicated RAM12GB ECC
  • Disk space250GB
  • TrafficUnlimited
  • IP address1
  • Claude Code, Codex, and GrokPreinstalled
  • Snapshot backupOne per month
  • BASIC managementIncluded
  • Setup feeFree
  • Server locationIsrael
  • Operating systemA range of Linux distributions, including AlmaLinux, Rocky Linux, Ubuntu, and Debian
  • FirewallIncluded
  • Graphs and monitoringIncluded

ENTERPRISE

  • CPU cores8
  • Dedicated RAM16GB ECC
  • Disk space300GB
  • TrafficUnlimited
  • IP address1
  • Claude Code, Codex, and GrokPreinstalled
  • Snapshot backupOne per month
  • BASIC managementIncluded
  • Setup feeFree
  • Server locationIsrael
  • Operating systemA range of Linux distributions, including AlmaLinux, Rocky Linux, Ubuntu, and Debian
  • FirewallIncluded
  • Graphs and monitoringIncluded

PREMIUM

  • CPU cores8
  • Dedicated RAM32GB ECC
  • Disk space400GB
  • TrafficUnlimited
  • IP address1
  • Claude Code, Codex, and GrokPreinstalled
  • Snapshot backupOne per month
  • BASIC managementIncluded
  • Setup feeFree
  • Server locationIsrael
  • Operating systemA range of Linux distributions, including AlmaLinux, Rocky Linux, Ubuntu, and Debian
  • FirewallIncluded
  • Graphs and monitoringIncluded

What is installed on your development server

The server comes ready to work: the Claude Code, Codex, and Grok coding agents are preinstalled, along with a configured firewall and the operating system you chose. The components below show how the server is built, from your code down to the infrastructure, and each one is marked as yours or maintained by us.

Yours

  1. Your projects and codeYou develop the projects, choose their programming languages, runtimes, and dependencies, and manage the code in Git repositories.
  2. The agents' API keys and permissionsYou connect the coding agents to the AI providers' models with API keys, billed directly by the provider, and decide which actions the agents may take without your approval.
  3. The Claude Code, Codex, and Grok coding agentsClaude Code, Codex, Grok, and similar tools are installed and configured when the server is set up, and from then on they are fully under your control.

Looked after by us

  1. FirewallFirewall rules are preconfigured on the server.
  2. Operating systemWhen you order, you choose from a range of Linux distributions, including AlmaLinux, Rocky Linux, Ubuntu, and Debian, and we install the operating system updates. You also get full root access over SSH.
  3. Virtual server with guaranteed resourcesThe CPU cores and ECC memory are guaranteed to your server alone, and its disk space is allocated on high-speed storage.
  4. Infrastructure, network, and backupThe server runs on infrastructure with dual power supplies and high bandwidth, gets a dedicated IPv4 address, and has DDoS protection active on it. A Snapshot backup is created for it automatically at the start of every month.

From your order to the agent's first task

We set up the server and install the coding agents, so you go straight from choosing a plan to working on your code, with no installations and no server configuration.

  1. Choose a plan and an operating system

    Choose the plan that fits the scope of your projects, and the operating system from a range of Linux distributions, including AlmaLinux, Rocky Linux, Ubuntu, and Debian.

  2. The server is set up with the coding agents installed

    Once your order is confirmed, our team sets up the server and installs Claude Code, Codex, and Grok on it, along with a preconfigured firewall. When setup is complete, you receive your access details for the server.

  3. Connect over SSH and bring in your code

    Connect to the server from your terminal or your IDE, such as Visual Studio Code, and clone your projects' code repositories to the server.

  4. Give the agent its first task

    Enter your AI provider's API key, start the agent in the project folder, and describe the task, such as fixing a bug or adding tests.

How to work with coding agents on your own server

A coding agent reads code, runs commands, and changes files, and on your own server it does this in an environment separate from your personal computer, at any hour and with guaranteed resources. Here are six common ways to use an AI coding server.

  • Remote development over SSH

    Connect to the server from Visual Studio Code, Cursor, or JetBrains IDEs, and edit, run, and test your code on the server from any computer.

  • Long tasks for the agent

    Give the agent a long task, such as upgrading dependencies or writing tests for the whole project, and it keeps working on the server even after you close your computer.

  • An isolated environment for the agent

    The agent runs commands and installs packages on the server, not on your personal computer, so the files, passwords, and accounts on your computer stay out of its reach.

  • Several agents in parallel

    Each agent works on its own Git branch, on a different task or a different project, and you review and merge the changes when they are done.

  • Tests and builds on the server

    The agent runs the project's tests and build process on the server's cores, fixes what fails, and runs them again.

  • A development server for your team

    Each developer gets their own user on the server, with their own SSH and API keys, and the whole team works in a consistent environment with the same tool versions.

Why develop with AI on an ErezNet server

We are a leader in cloud services, and over the years we have gained extensive experience working with business customers in a wide range of fields. Our development servers run on infrastructure with ECC memory and dual power supplies, and we maintain the server as described on this page, so you can focus on development.

  • Server maintenance by our team

    Our team installs and configures the coding agents when the server is set up, and from then on maintains the server and updates the operating system on an ongoing basis. Your time goes into development, not into server maintenance.

  • Guaranteed resources

    The server runs on multi-core Intel processors, ECC memory, and high bandwidth, and your plan's cores and memory are guaranteed to your server alone, available to your agents, tests, and builds at any hour. As your projects grow, you move to a larger plan, and your code, tools, and settings stay in place.

  • Isolation, security, and backup

    The agents run on a separate server, isolated from your personal computer and from other customers' servers, and DDoS protection is active on the server. A Snapshot backup of the entire server, including your projects and settings, is created automatically at the start of every month.

  • Personal support and free migration

    Our support team is available in our HelpDesk system most hours of the day for any question about the server and the infrastructure. If your development environment currently runs with another provider, we will move it to the new server at no cost.

Technologies and vendors we work with

  • Dell
  • HPE
  • Juniper
  • Fortinet
  • pfSense
  • OPNsense
  • Cisco
  • Check Point
  • Citrix
  • CloudLinux
  • Microsoft
  • Intel
  • LiteSpeed
  • DirectAdmin
  • AlmaLinux
  • Cloudflare
  • Node.js
  • Docker
  • MariaDB
  • MySQL
  • NGINX
  • cPanel
  • Plesk

Free migration of your development environment from another provider

Our team moves your development environment from your current server to the new server at no cost, along with its projects and files. The migration is scheduled with you, and you can keep working on the previous server until the switch.

  • Work on the previous server continues as usual throughout the move
  • Your projects and files are copied to the new server
  • The switch to the new server happens at the time we agreed with you

AI coding VPS FAQ

Before ordering an AI coding server, it helps to know what is installed on it, who maintains it, and how to connect to it. The answers also cover working safely with coding agents, API keys and secrets, Git, Docker and MCP servers, backups, and migration from another provider.

What is an AI coding VPS, and who is it for?

An AI coding VPS is a virtual private server that works as an independent computer with its own operating system and resources, with the Claude Code, Codex, and Grok coding agents preinstalled. A coding agent is an AI tool that runs in the command line: it reads your project's code, runs commands and tests, and makes changes to files, based on tasks you describe to it in natural language.

The server suits developers, development teams, and businesses that build software with AI and prefer the agent to work in an environment separate from their personal computer, at any hour and with guaranteed resources. Installing the coding agents, maintaining the server, and the monthly backup are handled by our team.

Why run a coding agent on a server rather than on my personal computer?

A coding agent runs commands, installs packages, and changes files, so it is best to run it in an environment separate from the computer that holds your personal files, passwords, and accounts. On your own server, the agent works only on what you have moved to it, and even if it takes a wrong action, your personal computer is not affected.

More advantages of your own server:

  • Long tasks keep running on the server even when your computer is off or offline.
  • Dedicated cores and memory for running tests and builds, without slowing down your personal computer.
  • The same work environment from any computer, and for every team member when needed.
  • A monthly Snapshot backup of the entire server, included on every plan.
Are Claude Code, Codex, and Grok preinstalled on the server, and what do I need to set up myself?

Claude Code by Anthropic, Codex by OpenAI, Grok by xAI, and similar tools are installed on the server when it is set up, along with firewall rules and the operating system you chose from a range of Linux distributions, including AlmaLinux, Rocky Linux, Ubuntu, and Debian.

After setup, all that is left is to connect the agents to your accounts and prepare your work environment:

  • An API key from each agent's AI provider: from Anthropic for Claude Code, from OpenAI for Codex, and from xAI for Grok.
  • A dedicated user for your work, with its own SSH key.
  • Your projects' code repositories, and the runtimes and tools they need.
Who maintains the server, and who updates the coding agents?

Our team installs and configures Claude Code, Codex, and Grok when the server is set up, and from then on maintains the server and updates the operating system throughout your service. Maintenance is included on every plan.

The coding agents and your projects are fully under your control: you guide the agents, write the code, choose the tools and dependencies of each project, and decide when to update the coding agents, which release new versions often. For questions about the server and the infrastructure, our support team is available in our HelpDesk system most hours of the day.

Which AI models do the coding agents use, and who pays for using them?

Claude Code works with Anthropic's models (Claude), Codex with OpenAI's models, and Grok with xAI's models. You connect the agents to the models with an API key you get from the provider, and usage is billed directly by the AI provider, based on how much you use, and is not included in the server price.

The models themselves run on the AI providers' infrastructure, so the server does not need a graphics processor (GPU): it runs the agent, the tools, and your projects. An agent working on a long task sends many requests to the model, so we recommend tracking your costs in your account with the provider and setting a spending limit there, where the provider allows it.

For no-code AI automations, an n8n VPS is a good fit, and for a personal AI agent available on WhatsApp and Telegram, an OpenClaw VPS.

Can I work on the server from Visual Studio Code and other IDEs?

Popular IDEs connect to the server over SSH and open your project folders on it directly: Visual Studio Code and Cursor with the Remote - SSH extension, and JetBrains IDEs, such as IntelliJ IDEA and PyCharm, with JetBrains Gateway. You edit in the familiar interface on your computer, while the files, the terminal, and code execution stay on the server.

In your IDE's built-in terminal, you can run the coding agents and follow the changes the agent makes, file by file. You can also work over a regular SSH connection, from any terminal app on Windows, macOS, or Linux.

Does the agent keep working after I close my computer or disconnect?

The agent's tasks keep running on the server even after you close your computer: start the agent inside tmux or screen, tools you can install on the server that keep your terminal session active even when you are not connected. The next time you connect, you return to the same session and keep following the agent from where it is.

Claude Code and Codex also have a non-interactive mode, which lets you run the agent from a script or a scheduled task (cron) and get the result when the task is done. For example, you can give the agent a long task in the evening, such as writing tests for every module of the project, and check the result in the morning.

How do I run a coding agent safely?

A coding agent runs with the permissions of the user who started it, so the main safeguard is to limit what that user is allowed to do. We recommend that you:

  • Create a dedicated user for working with the agents, without root privileges, and do not run the agents as root.
  • Keep the agent's approval mechanism on, so it asks for your approval before running commands and changing files, and pre-approve only routine actions, such as running tests.
  • Use modes that bypass approval prompts only in an isolated environment, and on a project whose code is backed up in Git.
  • Review the agent's changes before merging them into the main branch or deploying them to production.
  • Do not keep production passwords and keys on the development server.

In addition, content the agent reads, such as a web page, package documentation, or a bug report, may contain malicious instructions designed to make it act against your wishes. This risk is called prompt injection, and limiting permissions reduces the potential damage.

Can I set what the agent is allowed to do in each project?

Coding agents let you set their limits in advance: in Claude Code, you define in the project or user settings file which actions are allowed without approval and which are denied, and in Codex, you choose an approval mode and restrict access to files and the network. For example, in Claude Code you can pre-approve running tests, require approval for every package installation, and prevent the agent from reading secret files.

The agents also read an instructions file in the project folder, such as CLAUDE.md in Claude Code and AGENTS.md in Codex. In this file, you describe the project structure, the commands for building and running tests, and the code style rules, and the agent follows them in every task.

Can I run several agents in parallel?

You can run several agents on the server in parallel, each in its own terminal session, on the same project or on different projects. When several agents work on the same code repository, we recommend giving each one a separate working copy with git worktree, so each agent works on its own branch and the changes do not conflict.

Each agent, and especially the tests and builds it runs, uses cores and memory. To work with several agents in parallel, we recommend choosing a plan with more cores and memory, and tracking usage in the monitoring graphs.

Can I work on the server with Git, GitHub, and GitLab?

You can clone code repositories from GitHub, GitLab, or any other Git service to the server and work with them as usual: create branches, commit, and push your changes. The coding agents work with Git, and you can ask them to create a branch for each task, describe the changes in a commit message, and prepare the code for a pull request.

For access to private repositories, we recommend a dedicated SSH key for the server, or a deploy key for a single repository, rather than your personal key. We recommend reviewing the agent's changes in the pull request, like any code change, before merging them into the main branch.

Where do I store API keys, passwords, and environment variables?

The agents read the API key from an environment variable, such as ANTHROPIC_API_KEY in Claude Code, or from login details saved in the agent's settings. You set environment variables in your user settings on the server, and keep each project's secrets in the .env file in the project folder, which is not included in the Git repository.

To keep your keys safe, we recommend that you:

  • Restrict read permissions on secret files to the user who works on the project.
  • Configure the agent not to read secret files when it does not need them.
  • Create dedicated API keys for the server, with a spending limit, so you can revoke them at any time without affecting other systems.
  • Do not paste keys and passwords into your conversation with the agent.
  • Replace an exposed key right away, and revoke the old key with the provider.
Can I connect MCP servers to the agents?

Claude Code and Codex support MCP (Model Context Protocol), which connects AI agents to tools and data sources. An MCP server gives the agent more capabilities, such as access to a database, a task management system, up-to-date library documentation, or a browser for testing interfaces, and the agent uses them while working on the task.

MCP servers run on the server itself or as an external service the agent connects to, and are configured in each agent's settings. Third-party MCP servers are developed by outside parties, so the choice of which ones to use is yours. We recommend installing MCP servers only from trusted sources, and giving each one only the permissions and keys it needs.

Can I install Docker, runtimes, and databases on the server?

Root access lets you install the tools your projects need on the server: runtimes, such as Node.js, Python, PHP, and Go, databases, such as PostgreSQL, MySQL, and Redis, and Docker for running services in containers. You can also install more coding agents, such as Google's Gemini CLI, alongside Claude Code, Codex, and Grok.

We recommend installing tools from their official sources and updating them regularly. If your projects are all container-based, a Docker VPS is a good fit, with Docker and Docker Compose preinstalled and configured.

Can I view my project in a browser during development?

An application you are developing, for example a site built with Next.js or Django, runs on the server in development mode, and you can open it in the browser on your computer through an SSH tunnel, which encrypts the connection without exposing the application to the internet. Visual Studio Code, for example, creates such a tunnel automatically when it is connected to the server over SSH.

To show a test version to clients or team members, you can point a domain or subdomain to the server's IP address and issue it a Let's Encrypt SSL certificate, which renews automatically. We recommend protecting test versions with a password, so only people who received the access details can view them.

Does the server include full root access over SSH?

Full root access over SSH is included on every plan, alongside the ongoing server maintenance we provide. With this access, you choose versions, install packages, runtimes, and development tools, create users, view system logs, and configure every detail of your work environment as needed. You can also log in with SSH keys, which protect access better than a password.

We recommend using root for administrative tasks only, and working on projects and with the agents under a dedicated user. We recommend coordinating changes to operating system settings, such as firewall rules, with our team in advance, so operating system updates continue as usual.

Can several developers work on the same server?

You can create a separate user on the server for each developer, with their own SSH key and home folder, so each one works on their own projects and connects the agents to their own API key. All developers work in a consistent environment, with the same versions of runtimes and tools, and the server's monthly price does not depend on the number of users.

Since the users share the server's cores and memory, we recommend choosing a plan based on the number of developers and agents working in parallel, and setting file permissions carefully, so one developer cannot read another developer's secrets.

What affects the server's resource usage?

The AI providers' models run on the providers' infrastructure, so the agent itself uses moderate resources. Most of the server load comes from the actions the agent runs, such as installing dependencies, building the project, and running tests, from the services running alongside it, such as databases and containers, and from the number of agents and developers working in parallel.

In the monitoring graphs in our HelpDesk system, you can see when the server is approaching full CPU or memory usage, and upgrade it to a larger plan in good time.

Can I upgrade the server as my projects grow?

You can upgrade the server to a larger plan at any time, with more cores, memory, and disk space, and increase bandwidth to suit your business needs, by opening a ticket in our HelpDesk system. The upgrade needs only a short reboot, after which the server is back up with your projects, tools, and settings. We recommend scheduling the upgrade for a time when no agent tasks are running on the server, since the reboot stops them.

On request, you can also reduce cores and memory, but disk space cannot be reduced. For work that needs a different combination of resources than the plans offer, you can order a custom configuration by opening a ticket in our HelpDesk system.

Which operating systems can I choose for the server?

When you order, you choose from a range of Linux distributions, including AlmaLinux, Rocky Linux, Ubuntu, and Debian. Claude Code, Codex, and Grok are installed on each of them, and the agents work the same way.

We install and update the operating system, so you need no prior Linux administration knowledge. If you have no preference, choose AlmaLinux, the default option in the order form. Developers who are used to Ubuntu can choose it and work in a familiar environment.

Will you migrate my development environment from another provider?

Our team will move your development environment from your current server to the new server at no cost, along with its projects and files. Work on the previous server continues until the switch, and the switch date is set together with you.

To arrange the migration, open a ticket in our HelpDesk system.

Do you guarantee server availability?

Yes. We do everything we can to keep our servers available at all times, and we commit to at least 99% availability on an annual average, from the first day of service.

Is there a commitment, and do prices include VAT?

The service is billed monthly, with no commitment beyond the current month of service, no fine print and no cancellation fees. All plan prices include VAT.

Shopping cart

Your cart is emptyYou can start with one of our popular plans.

Site search

Accessibility

Text size
100%
Accessibility statement