# Deployment and models

Volobox runs where your organization needs it: on Volobox Cloud, or in your own private cloud or on-premise. Its AI works from the first day on models included with your workspace, or on model providers you connect with your own keys. This page is for the people who decide where Volobox runs and which models it uses; plans, prices and what each plan includes are on [Pricing](/pricing).

## Deployment options

- **Volobox Cloud.** We set up and run your workspace. An administrator chooses an update window for new releases, or upgrades when it suits the team; mandatory releases are installed either way. Microsoft 365 features, including Microsoft Entra ID sign-in, connect through one Microsoft 365 connection with nothing to register in Entra.
- **Private cloud or on-premise.** Volobox runs in infrastructure you control (which plans offer it is on [Pricing](/pricing)), and whoever operates it decides when it upgrades. Microsoft Entra ID sign-in uses your own app registration.

> **Note:** On-premise and private cloud start with a conversation rather than a trial: [request a demo](/demo).

## Models: included, or bring your own

A Volobox Cloud workspace arrives with model access we manage, so chat, agents and coding work on the first day with no provider setup. Administrators choose which of those models are enabled and which one coding sessions start on.

You can also bring your own models, with your own keys. Connect providers such as OpenAI, Anthropic and OpenRouter, Azure OpenAI deployments, or another OpenAI-compatible endpoint. Examples are model families such as Claude, GPT, Gemini and Kimi, and self-hosted open-source models. See [Add a model provider](/docs/add-model-provider).

> **Note:** Whether your plan includes your own providers is listed on [Pricing](/pricing). If a plan does not include them, a provider added earlier is paused with its settings kept, never deleted.

![Organization Settings, AI tab, Models: the LLM providers list with Anthropic, OpenAI and a turned-off Azure OpenAI deployment, each with its address, model catalog mode, whether it is enabled and how many of its models are enabled, and the Add provider button.](https://volobox.io/docs/images/ai-providers.light.webp)

*The LLM providers list in Organization Settings, AI, Models: each provider with its base URL, model catalog, whether it is enabled and how many of its models are on.*

## Coding agents

Coding agents add one more choice:

- **OpenCode** runs on the organization's models.
- **Cursor** and **Codex** run on each person's own account.
- **Claude Code** runs only on a person's own machine signed in to their Claude subscription, never in the cloud sandbox.

See [Coding](/docs/coding).

## Where sandboxes run

Agents run code, and coding sessions run, in **sandboxes**: isolated computers created for the work.

- **Cloud sandboxes** are the default.
- **Your machines**: a person can add a computer they own. It connects to Volobox outbound with no open port, and runs only that person's private work. See [Run coding sessions on your own machine](/docs/own-machine).

## Where your data lives

Data location follows the deployment you choose. On-premise or private-cloud deployments let you keep Volobox and your model endpoints in infrastructure you control.

Volobox sends the context a request needs to the model endpoints your workspace configures. Whether a provider retains requests or uses them for training depends on your agreement with that provider.

For managed-cloud regions, retention and backup requirements, bring your policy to the security review: see [Trust](/trust).

## Related

- [Security and administration](/docs/security-and-administration)
- [Coding](/docs/coding)
- [Integrations](/docs/integrations)
- [Agents](/docs/agents)
- [Pricing](/pricing)
- [Trust](/trust)
