AI Governance (Unified LLM Int...
Configure Unified LLM Interfac...
Virtual Keys
6 min
virtual keys are used to provide api access to the llms, models, and providers you have configured in amberflo a virtual key is always tied to a workload this means a key only grants access to the models that have been explicitly allowed for that workload virtual keys make it possible to tightly control who can access which providers and models without ever sharing direct provider api keys all access to llms through the amberflo gateway happens via virtual keys why virtual keys matter virtual keys allow you to grant scoped access to llms without exposing provider credentials enforce model level access through workloads rotate or revoke access without impacting other consumers automatically attribute usage and cost to the correct workload every request made with a virtual key is authenticated, governed, and attributed by the gateway viewing existing keys to manage virtual keys go to access management in the left hand navigation select the keys tab this page displays all previously created keys, including key name a partial key value (for identification only) associated workload creation date last updated date from this view, you can also delete keys deleting keys deleting a key immediately invalidates it any application or service attempting to use the deleted key will receive an unauthorized error when making requests through the gateway deleting a key does not affect the underlying workload, models, or provider credentials creating a virtual key to create a new key click add key in the upper right enter a key name the name is for your reference and should clearly indicate the purpose of the key, such as the application, service, or environment it is used for select a workload a key must always be attached to a workload to ensure proper access control and attribution if you have many workloads, you can start typing in the search field to filter the list click create key saving the key value once the key is created, the full key value is shown one time only copy the key and store it securely you will not be able to view the full key value again if the key is lost, you must create a new one after copying the key, click done to return to the keys list the new key will now appear in the list using virtual keys virtual keys are used exclusively for accessing the amberflo ai gateway they are not the same as an amberflo platform api key use the virtual key in your api requests when calling models through the gateway as soon as requests are made, usage and cost data will begin flowing and will be automatically attributed to the associated workload what’s next with a virtual key created, you are ready to make your first api call through the gateway refer to the api usage examples to see how to call models using a virtual key and a unified openai compatible interface
