The PhariaOS API
The PhariaOS API is a powerful resource management system that oversees the entire lifecycle of resources, from inception to maintenance, while efficiently integrating them with PhariaOS. Additionally, this program takes charge of seamless application deployments, ensuring smooth operations throughout.
See also The PhariaOS operations manual.
Authentication
To use the API, you need an authentication token. You create these in PhariaOS. See Managing credentials and tokens.
Endpoints
The API endpoints cover most of the PhariaOS functionality:
| Endpoint | Description |
|---|---|
|
Use this endpoint to manage applications and their deployments. You can create, update, and delete an application, list all applications, and retrieve an application by its ID. You can also deploy and undeploy applications, list all deployments of an application, and retrieve an application deployment by its ID. |
|
Use this endpoint to retrieve the available GPU resources for a cluster, list all nodes in a cluster, retrieve the taints of a cluster node, and retrieve a cluster node by its name. |
|
Use this endpoint to retrieve a list of all available models that were installed using Helm charts. A model is considered 'available' if it has at least one worker that is accepting requests. The list of models is provided by the PhariaInference API endpoint |
|
Use this endpoint to manage models that were deployed using PhariaOS. You can list all available foundation models as well as all available inference runtimes supported to deploy models. (Currently, the supported inference runtimes are Aleph Alpha authorial Luminous and vLLM.) In addition, you can create, update, and delete a model, list all available models, and retrieve a model by its ID. A model is considered 'available' if it has at least one worker that is accepting requests; models that are still deploying or failed to deploy are considered unavailable. |