Frequently Asked Questions
Common questions about how weightroom works, billing, and security.
Private AI setup
What is weightroom?
weightroom helps businesses run private AI for their software. Assisted setup is available today: weightroom operates the AI deployment, and your developer or IT partner connects it to your tools. GPU Workspaces are a separate self-service product for people who want to rent a GPU and manage their own development or experiments.
Can I launch my private AI deployment instantly?
No. Private AI setup is assisted today. Tell us the business task and we will follow up by email to discuss the configuration, data handling, and price before work begins. The self-service GPU Workspace flow does not launch a managed private AI deployment.
How do I get help with setup?
Use the Get setup help form to describe the task and include your developer or IT partner if you have one. We will follow up by email. No deployment starts until the setup and price are agreed.
GPU Workspaces
These answers describe the separate self-service GPU product, not an operated private AI deployment.
Do you store or have access to my API keys?
For GPU Workspaces, configuration and credentials used to run a service may be stored encrypted at rest and passed to the infrastructure provider running your instance. Tools you enable can read those credentials and may send data to their own services. weightroom and infrastructure providers can have technical access for operations, support, security, or legal obligations. Do not include credentials in a setup request or support message.
Can you see what's running inside my instance?
A GPU Workspace runs in an isolated environment, but isolation does not mean weightroom or its infrastructure providers are technically unable to access data. Enabled features such as file access and backups process workspace content. We also process operational metadata such as instance status, uptime, and resource type.
How does billing work?
GPU Workspace compute is billed per minute from when a GPU is reserved, before the environment is fully ready. The rate shown at launch is set for that instance session. Credits are deducted automatically. After an instance has stopped, GPU compute billing stops; attached storage and separately retained backups can continue to incur charges. Destroying the primary instance does not necessarily delete separate volumes or backups.
Can I get a refund for unused credits?
All credit purchases are final and non-refundable. Credits remain in your account until used and do not expire. If you experience a platform issue that caused unintended charges, contact [email protected] and we'll review your case.
What happens to my work when I stop an instance?
Once a GPU Workspace has stopped, GPU compute charges stop. Attached storage can keep accruing charges. Backups use separate storage and may outlive the original instance; a backup-on-destroy setting can create a final copy. Destroying the instance removes its primary files, but you must delete separate volumes and backups or wait for their retention period to end. Review the storage and backup details before relying on deletion for a data-retention deadline.
What GPU options are available?
We offer GPUs including RTX 3090, RTX 4090, A100 80GB, and H100 SXM through our provider network. Availability and pricing depend on real-time supply and can change. After a launch attempt, review your account for its final status and billing; contact support about unexpected charges.
Do I need my own GPU?
No. We provision remote GPUs from cloud providers. All you need is a browser.
What if a template uses both GPU and LLM API calls?
GPU workspaces are billed by weightroom. If you configure your workspace to call an external API, that API provider bills those calls separately. Monitor external API usage in your provider account.
Can I use my own datasets?
Yes. Datasets on the launch page are optional pre-loaded options to get you started quickly. You can skip them and upload or download your own data once JupyterLab is open.
How long does it take to launch an instance?
Launch times depend on GPU availability and container startup. Billing begins when a GPU is reserved, before the environment is fully ready. If a start fails, review your account for its final status and billing, and contact support about unexpected charges.
What happens if I run out of credits?
You'll receive a low-balance warning so you can add credits before any interruption. When your balance reaches zero, running instances are automatically stopped. In rare cases, instances may be interrupted due to infrastructure issues beyond our control.
Account
Who can use weightroom?
You must be at least 18 years old to create an account and use weightroom. By signing up, you agree to our Terms of Service and Privacy Policy.
Have a question not covered here? Reach out at [email protected]