6.3 KiB
title, description
| title | description |
|---|---|
| Models and Providers | Choose an account, connect a service or add a model without changing the video execution system. |
A Model defines what you ask to generate: its inputs, supported parameters and output type. A Provider knows how to fulfill that request through a particular service. An Endpoint is a configured instance of that Provider, with its service address, credential reference and capacity. The Runtime Profile binds the requested capability to an Endpoint.
The starter Profile includes local services and HypiHub. Other services connect through a Provider package installed by the video project. Tell the Agent which service you want and provide its API documentation; the Agent can install an available package or create a project Provider. Model and deployment services introduces service choices. Choosing a service is separate from choosing an Agent environment.
HypiHub is the recommended integrated hosted service. BYOK means bringing an account's own API key: the key connects to the service that issued it, through a compatible Provider. Tell the Agent which service you already use and provide its API documentation; the Agent can handle the connection and project package. A key authorizes requests but does not implement that API. One project can use different services for different capabilities, with each service's own account and charges.
Choose the change that matches the need
| You want to… | Change |
|---|---|
| Use another key for the same service | The credential reference and selected Endpoint configuration |
| Use another compatible service address | The address or deployment options supported by the Provider |
| Run the model on your own cloud deployment | Prepare its inference service, then configure a compatible Provider or implement its API |
| Use the same model through a different API | Install or write a Provider for that API and select its Endpoint |
| Use a model not yet defined | Add a Model package and a Provider that supports its request |
Two services offering the same model can have different request formats, limits or available parameters. The Provider checks the request against that service's support and explains a mismatch. A Profile chooses the route; an error on that route does not authorize spending through another account.
For an existing installation, inspect the selected Profile and credential status first. A starter Profile supplies configuration examples; choose the services you want before connecting accounts or preparing their dependencies. Invocations and Builds shows the commands.
Use your own model deployment
You can run a model on compute you control and connect the resulting inference service to Hypit. The cloud platform provides deployment and compute; the Model defines the generation request; the Provider implements the serving API; the Endpoint selects the deployed address and credentials.
An existing Provider can connect to a deployment with the same complete protocol. A different API
can use a project package such as packages/provider-my-cloud/. The serving protocol determines
that implementation boundary, rather than the cloud platform's brand. If the model is not already
described by Hypit, add its Model definition as well.
Preparing a deployment can involve model files, compute configuration and a serving process. Your Agent can use the chosen platform's and model's instructions to establish that service, its persistence and compute costs. Platform-management access and inference access may be separate. Once available, the deployment receives ordinary requests through its Endpoint; provisioning a deployment and executing a video Build have separate lifetimes.
Add a Model
Create a project package when the service introduces a request shape that the installed Models do not describe. Define the prompt, reference roles, supported choices and output media from the author's point of view. The Generation package README provides the exact API and a package example for the selected release. Credentials and HTTP mapping belong to the Provider.
Add a Provider
Create a project Provider package to connect a selected Model to an API or local service. Implement
the request forms and limits that service actually supports, install the package with the project,
then configure it under endpoints and select it through bindings in the
Runtime Profile. The Endpoint package README
owns the exact TypeScript API and package fields.
The complete project Provider example
demonstrates reference uploads, task receipts, collection and pricing using an illustrative API.
It also ships with the executable, so the Agent can adapt it without a repository checkout. It holds
two packages: provider-images for a generated image, and provider-videos for a generated video
whose service receives the wider reference vocabulary — images, videos, audio and first/last frames —
and returns its result through a separate collection step.
Implement the capability that matches your service's request shape; a service that renders both
images and video can declare both capabilities in one package. A service often supports a narrower
range than the Model's vocabulary allows, such as fewer resolutions or a lower maximum duration.
That difference belongs to the Provider: report it from the capability's supports so plan refuses
the request with a reason, rather than editing the shared Model or silently narrowing the author's
request.
Prices and permission
The Provider declares local work without a Provider charge, or supplies its published pricing page.
It can also read current rates using the Endpoint's credentials and return a concise summary with
the original pricing documents. hypit pricing <source> brings those rates together with the planned
requests. Pending media measurements remain unknown until the material exists.
Rates help explain the cost. The user's agreement supplies permission to spend through the selected account for the agreed work and budget. That permission is separate from a successful login or an available balance.