From your first model
to your next renewal.

Slab brings image preparation, camera-frame inputs, model execution and task-specific output helpers into one SDK. It also handles hardware backend selection and includes model protection. Your team integrates that pipeline through native language bindings.See a complete integration ↗

Yes. Convert your ONNX model into a Slab pack with the conversion tool. It reports unsupported operators during conversion. Then run the pack on your target devices to check the backend, output quality and performance for your application.Convert your model ↗

Start with the public model gallery. Choose a task, review its input and output examples, and download a model pack. The classification example on the developer page gives you a concrete first integration.Browse the model gallery ↗

Slab supports iOS and iPadOS 18+, Android 12+ on arm64, and macOS 15+ on Apple silicon. Use Swift, Kotlin, Java, C or Python according to your platform. The setup guides list the package and hardware requirements for each.Check platform compatibility ↗

Slab checks compatibility between the model and the device before selecting a backend. GPU and NPU availability depend on the model and hardware. You can request a specific backend and benchmark it during the trial. If a result is unexpected, share the details with our support team.Understand backend selection ↗

Create an account, register the application you want to evaluate, and use its trial license with the SDK. Choose a gallery model or convert your own. The trial gives your team 30 days to evaluate on your devices before purchasing. No credit card is required.Start your trial ↗

Yes. Report an issue from your dashboard during the trial. Share the conversion error or unexpected result and the device you are testing. We can help distinguish an integration problem, an SDK defect and a capability your project would need added.Ask the Slab team ↗

An annual Slab runtime license for one product, unlimited installations, built-in model protection, and a full year of integration guidance and SDK bug support. Slab has no per-user or per-inference fees.See pricing ↗

The license covers one product. Development builds, testing variants and flavors of that product belong together. Unrelated products require separate licenses. Each license is tied to the registered application identity.Read the licensing guide ↗

Yes. Model protection is part of Slab and included in the annual license. You can evaluate it with your proprietary model during the trial. Use your intended deployment configuration when measuring performance.

The public gallery provides model downloads for use with Slab. Each model retains its own license, shown on its gallery page. Your Slab license covers the runtime; use of a model follows that model’s terms. You can also run a model you own.Explore model licenses ↗

Yes. Slab runs inside your product. Your customers use your interface and licensing experience, and do not need an imaged account. Your team manages the Slab license and provides it to the application.

Slab is available as a standalone SDK with an annual license. Your developers can integrate it using the packages, examples and documentation. Custom model training or adding new capabilities can be scoped separately if your project needs them.Open the developer guide ↗

Request the annual license from your application’s page in the dashboard. The team handles the invoice, and your issued license is available to copy or download there. Your integration continues with the production license.Open your dashboard ↗

The report goes to the Slab team. We investigate the issue, ask for any details needed to reproduce it, and share progress in your dashboard. Production blockers, crashes and incorrect results take priority. We share available workarounds and identify the release containing a fix.Report an issue ↗

Training the model yourself does not exclude you from Slab support. We investigate defects in supported SDK functionality. Successful conversion is the first check; backend compatibility is also checked on the device. Share the model’s conversion result and the failure so we can establish whether the SDK or an unsupported capability is involved.Get help with your integration ↗

We prioritize by impact and keep the investigation in your support thread. The standard license does not include a contractual response or resolution deadline. If your application needs a specific service commitment, discuss it with us before purchasing so expectations are agreed in writing.Discuss your support requirements ↗

Inference and runtime license verification happen on the device. Your application can keep running offline with a valid license. Plan how it will receive a renewed license before the current one expires.

Arrange renewal ahead of expiry and distribute the renewed license while devices are connected. Your existing app configuration or update process can carry it. If your fleet has long offline periods, contact us early to coordinate issuance and leave time to distribute and test the new license.Plan license renewal ↗

An expired license prevents new inference runs; a run already in progress finishes. Provide a renewed license to restore operation. Distributing the renewal before expiry keeps your application’s AI feature available.License renewal and recovery ↗