> ## Documentation Index
> Fetch the complete documentation index at: https://docs.costgraph.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Tensorlake sandboxes

> Run CostGraph workflows and CI jobs in Tensorlake sandboxes: the key to create, its permissions, and its limits

Tensorlake Sandboxes are isolated MicroVM environments that start quickly and keep memory and filesystem state across suspend and resume.

The following table shows what CostGraph runs on this provider.

| Workflows | CI runners | Connect with |
| - | - | - |
| Yes | Yes | Integrations or your own dispatcher |

## Connect

Connect Tensorlake on Integrations, and CostGraph starts sandboxes in your Tensorlake account. You don't run anything yourself.

1. In CostGraph, go to **Integrations > Sandbox > Tensorlake**.

2. Enter these fields:

   * **API key** (`TENSORLAKE_API_KEY`, required): A Tensorlake project API key from the Tensorlake dashboard.

3. Click **Connect**.

For CI runners, CostGraph sets CPU and memory and boots `providerTemplate` as the Tensorlake image.

### Your own dispatcher

To keep the key out of CostGraph or reach a private network, run your own dispatcher instead.
Set `provider` to `tensorlake` and the preceding fields as environment variables, as described in [Run your own dispatcher](https://docs.costgraph.ai/costgraph/workflows/managed-sandboxes.md#run-your-own-dispatcher).

## Key and permissions

TENSORLAKE\_API\_KEY is a Tensorlake project API key (tl\_apiKey\_...) created in the Tensorlake Dashboard: select the project, then create an API key.

keys are not scoped: each key lives in one project with the same permissions as a project member, and cannot be restricted to specific endpoints or given org-level permissions. Source: [https://docs.tensorlake.ai/platform/authentication](https://docs.tensorlake.ai/platform/authentication)

Sources: [docs.tensorlake.ai: authentication](https://docs.tensorlake.ai/platform/authentication), [docs.tensorlake.ai: billing](https://docs.tensorlake.ai/platform/billing), [docs.tensorlake.ai: create](https://docs.tensorlake.ai/api-reference/v2/sandboxes/create), [docs.tensorlake.ai: introduction](https://docs.tensorlake.ai/sandboxes/introduction).

## Limits

* Max session duration: Free 2 h, Usage Credits 24 h, Pro and Enterprise unlimited; default timeout 600 s; unnamed sandboxes are discarded at timeout, named ones suspend. ([source](https://docs.tensorlake.ai/platform/billing))
* Per sandbox: Free 1 vCPU/1 GB; Usage Credits 4 vCPU/16 GB; Pro 16 vCPU/64 GB; root disk 10-100 GB; RAM 1-8 GB per vCPU on paid plans. ([source](https://docs.tensorlake.ai/platform/billing))
* Concurrent sandboxes: Free 1, Usage Credits up to 100, Pro 1,000, Enterprise unlimited. ([source](https://docs.tensorlake.ai/platform/billing))
* API timeout\_secs maxima: Free unverified 1 h, Free verified 2 h, On-Demand 24 h (0 = plan max). ([source](https://docs.tensorlake.ai/api-reference/v2/sandboxes/create))
* Regions and CPU architecture not documented; GPU sandboxes use gVisor, CPU-only run in VMs. ([source](https://docs.tensorlake.ai/sandboxes/introduction))

## Related

* [Supported sandbox providers](https://docs.costgraph.ai/costgraph/workflows/sandbox-providers.md)
* [Run workflows on a managed sandbox provider](https://docs.costgraph.ai/costgraph/workflows/managed-sandboxes.md)
* [Run CI jobs in sandboxes](https://docs.costgraph.ai/costgraph/autopilot/ci-sandboxes.md)


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.