On-device licensing is for teams that want BackgroundErase to run directly inside the app instead of calling a hosted API for every image. This is the right licensing path when background removal is meant to feel like a native mobile capability rather than a cloud round trip hidden behind the UI.
We currently offer ready-to-ship Core ML and TensorFlow Lite packages for this deployment model. In practice, that means iOS and Android teams can integrate the model into their own app experience and keep inference closer to the user.
Simple summary: on-device licensing lets your team ship BackgroundErase directly inside a mobile app using Core ML or TensorFlow Lite. It is usually priced above self-hosted licensing because local execution, mobile packaging, and premium app UX create a different kind of value.
What on-device licensing means
In an on-device deployment, the model is bundled into the application itself instead of being called remotely from a public server endpoint. The app loads the model locally, runs inference on the device, and continues the user workflow without needing a hosted API request for every operation.
That changes the product experience in a meaningful way. A mobile feature that runs locally often feels faster, more native, and more integrated into the rest of the app. For some teams, that is the whole point of choosing this licensing model.
Core ML and TensorFlow Lite packages
Our on-device licensing path centers on ready-to-deploy model packages for the two most common mobile directions:
- Core ML: for teams building native iOS experiences and wanting a production-ready Apple-side deployment path
- TensorFlow Lite: for teams targeting Android or multi-device mobile packaging through the TensorFlow Lite ecosystem
These packages are designed to help teams move faster than if they were trying to convert and productionize the model entirely on their own. The goal is to reduce friction between “we want this feature in our app” and “we have something stable enough to ship.”
Why teams choose on-device
Teams usually choose on-device licensing for one or more of the following reasons:
- Lower perceived latency for the end user
- A more native and responsive mobile UX
- Less dependence on a network request for each image
- Stronger local-processing story for privacy-conscious apps
- More predictable runtime economics without per-call API usage
- Better fit for products where image editing is a core feature
In many mobile products, the difference between local inference and cloud inference is not just technical. It is a product decision that shapes how polished the app feels.
Common motivation: teams want background removal to feel instant and native inside the app instead of feeling like a slow remote operation.
Best-fit use cases
On-device licensing is usually the strongest fit for mobile products where background removal is a visible feature rather than a hidden backend step. Common examples include:
- Photo and video editing apps
- Creator tools and social media apps
- Camera or capture workflows
- Ecommerce seller tools for product listings
- Marketplace listing apps
- Consumer mobile apps where speed and polish matter a lot
If the feature is meant to be experienced directly by the person holding the phone, on-device often makes more sense than routing everything through a backend.
Yearly and perpetual licensing
On-device licensing can be structured as either yearly or perpetual.
A yearly structure is often the better fit for teams that want more flexibility, cleaner annual budgeting, or an easier starting point before committing to a longer-term licensing path. A perpetual structure is often the better fit for teams that want a more durable long-term agreement around shipping the runtime in their app.
The right choice usually depends on how the mobile team budgets, how stable the product roadmap is, and whether the company wants the licensing conversation to behave more like an annual software relationship or a longer-term embedded technology investment.
Optional maintenance packages
Yearly and perpetual on-device licenses can also include an optional maintenance package. This is usually the right add-on for teams that want enterprise support, rollout guidance, help with future model updates, or a stronger technical relationship after the initial handoff.
A maintenance package can be especially useful in mobile deployments because app teams often care about update cadence, long-term compatibility, and having a clear support path when the model becomes part of a live user-facing product.
- Enterprise support access
- Guidance on rollout and packaging questions
- Help with updates and future revisions
- A stronger long-term deployment relationship
- Better support coverage for production apps
On-device versus self-hosted
The easiest way to compare the two licensing paths is to ask one question: where should inference happen?
- Choose on-device if you want the model to live inside the mobile app.
- Choose self-hosted if you want the model to live on your servers or cloud infrastructure.
If the feature is primarily about app-side user experience, on-device is usually the better fit. If it is primarily about backend control and server-side orchestration, self-hosted is usually the better fit.
Good fit for premium mobile experiences
On-device licensing is particularly strong when a company is trying to build a premium experience. Local inference can help the feature feel cleaner, faster, and more integrated into the rest of the product. That is why this option is common in apps where design polish and responsiveness matter as much as raw capability.
In those cases, the licensing decision is closely tied to the product strategy. The company is not just buying background removal. It is buying a better way to deliver that capability inside its own mobile experience.
Operational considerations
Even though on-device removes the hosted runtime from the critical path, it still requires careful product planning. Teams need to think about packaging, app updates, hardware compatibility, rollout strategy, and how the model should fit into the broader mobile architecture.
That is one reason many companies pair on-device licensing with maintenance and enterprise support. The model package is only one part of the story. The long-term deployment path is just as important.
Related licensing paths
If you are comparing deployment options, you may also want to review the general licensing overview and the self-hosted path before deciding which model fits your team best.
The simplest version
On-device licensing lets your team ship BackgroundErase directly inside an iOS or Android app using ready-to-go Core ML or TensorFlow Lite packages. It is generally priced above self-hosted licensing and can be structured as either yearly or perpetual, with optional maintenance for enterprise support.
Contact sales
If your team wants to ship BackgroundErase directly inside a mobile product, visit backgrounderase.com/enterprise and tell us whether you are targeting iOS, Android, or both.