Plans
Bring your own keys
Available now.
- What you supply
- Your own model-provider and web-search API keys, added in Valet's settings after installing.
- What we supply
- The Valet application, its updates and its signed plugins.
- How models are chosen
- Any model from any provider you hold a key for, hosted wherever you choose.
- How it is billed
- No charge from Noerion Labs. You pay the model provider directly for what you use.
What holds on this plan
- Bring your own keys: No part of your work reaches Noerion Labs.
- Nothing of your work is kept on Valet's servers.
- Bring your own keys: Valet talks directly to the model provider you configured. No request passes through Valet's servers.
- Every action has a risk level. High-risk actions wait for your approval, and no model can approve or skip them, by design.
- A prompt-injected or confused model can only propose.
- The risk table is in code, outside any model.
- Changing a risk level is a deliberate setting, never a chat reply, by design.
- New tools are compiled .NET code, checked and signed before they load.
- Any model, hosted anywhere you choose, including open models in UK, EU or Swiss data centres.
- The application sends no usage or telemetry data.
- Bring your own keys: Your provider keys stay on your PC and are never sent to Noerion Labs.
- Planned: The risk table, threat model and injection-test results will be published.
Pro
Planned. Not available yet, and there is nothing to buy or sign up for.
- What you supply
- No keys to manage.
- What we supply
- Access to the models on the menu, relayed through Valet's servers to the provider of the model you choose.
- How models are chosen
- From a menu, each model stating its provider, hosting jurisdiction and retention period.
- How it is billed
- One bill from Noerion Labs.
What holds on this plan
- Nothing of your work is kept on Valet's servers.
- Every action has a risk level. High-risk actions wait for your approval, and no model can approve or skip them, by design.
- A prompt-injected or confused model can only propose.
- The risk table is in code, outside any model.
- Changing a risk level is a deliberate setting, never a chat reply, by design.
- New tools are compiled .NET code, checked and signed before they load.
- The application sends no usage or telemetry data.
- Pro, planned: On the managed plan, requests will pass through Valet's servers to the model you chose, and will not be stored there.
- Planned: The risk table, threat model and injection-test results will be published.
For firms
Pro, offered to firms by the seat, is a direction we intend to take.