Download

Private, on-device AI for Android — Cove is in development.

Cove is developing focused tools for travel translation, voice notes, photo understanding, and health guidance. The project explores local Gemma inference; public app releases and end-to-end offline behavior are not yet verified.

Release updates See why offline

One Cove family. Four Android app concepts in development.

Cove's project explores on-device Gemma inference. A local model run does not by itself establish that every app feature is offline or that no data leaves the device.

Why offline matters

Three reasons your data — and your AI — should stay home.

  • Know what on-device AI does — and does not — mean

    On-device inference means a model processes its input on the device. It does not prove that an entire app never sends data elsewhere. Cove's Android tools are in development, and their complete public-release data flows have not yet been verified. Check the release privacy information before using sensitive content.

    Read the security whitepaper

  • Offline behavior needs a release check

    A locally installed model can perform inference without contacting a cloud model, but downloads and other app features can require internet. Cove's public apps are not yet available for an end-to-end airplane-mode check, so broad offline claims remain unverified.

    See the offline demo

  • Local inference avoids a model round trip

    When a task is processed by an on-device model, that inference does not need a network trip to a remote model. This does not guarantee a particular response time: results depend on the device, model, and task. Cove has not published verified performance benchmarks.

    Understand on-device inference timing

Pricing

Prices will be announced with the public release.

No public prices or purchase terms have been confirmed. Follow release updates for verified details.

Get release updates

Common questions

Five short answers about Cove’s current development and release status.

  • Does Cove really work without internet?

    Cove is developing Android tools that explore on-device Gemma inference. When a model is installed, local inference can process a prompt without sending that prompt to a cloud model. This does not prove every app feature works offline: downloads and other services may need a connection. There is no verified public app release yet, so full airplane-mode behavior cannot currently be checked.

  • Is Cove really free to try?

    Cove's public download page currently offers release notifications, not a verified app package. Trial limits, paid features, and prices have not been verified for a public release. Check the release information when a build becomes available.

  • Which AI model does Cove use?

    Cove's Android project includes a Gemma-based local inference path using LiteRT-LM. Model variants, download sizes, device compatibility, and quality comparisons are not verified as public release details. On-device inference describes where that model processes its input; it does not by itself establish the network behavior of an entire app.

  • Is Cove safer than ChatGPT for sensitive data?

    A local inference path can keep a prompt on the device for that model run, but that alone does not establish how every app feature handles data. Cove's public builds and end-to-end data flows are not yet available for verification, so we cannot make a blanket safety comparison. Review a release's permissions and privacy information before using sensitive material.

  • Why offline AI instead of cloud?

    On-device inference can avoid sending each prompt to a remote model and can work without a network once required files are installed. Other app features may still need connectivity, and actual speed depends on the device and workload. Cove is exploring these trade-offs; public release behavior has not yet been verified.

Read pre-release help