Unable to use model in the app

[MrMati]

Search before asking

• I have searched the HUB issues and found no similar bug report.

HUB Component

Export

Bug

Model details view has no export/download button and no add to fav button.

Environment

OnePlus Nord 4
OxygenOS 15 - Android 15
Ultralytics HUB 0.9.36

Minimal Reproducible Example

1. Have a trained and never used model in project
2. Directly go to predict, select it, confirm download
3. Nothing happens
4. Go to the model details through Projects tab
5. Model has no buttons

Additional

I have cleared cache of the app, nothing changed.

Everything for this model still works fine in the web app, I can export to a new format etc.

BTW The last update (9 minutes ago) was just a name change.

[UltralyticsAssistant]

👋 Hello @MrMati, thank you for raising an issue about Ultralytics HUB 🚀! Please visit our HUB Docs to learn more:

• Quickstart. Start training and deploying YOLO models with HUB in seconds.
• Datasets: Preparing and Uploading. Learn how to prepare and upload your datasets to HUB in YOLO format.
• Projects: Creating and Managing. Group your models into projects for improved organization.
• Models: Training and Exporting. Train YOLOv5 and YOLOv8 models on your custom datasets and export them to various formats for deployment.
• Integrations. Explore different integration options for your trained models, such as TensorFlow, ONNX, OpenVINO, CoreML, and PaddlePaddle.
• Ultralytics HUB App. Learn about the Ultralytics App for iOS and Android, which allows you to run models directly on your mobile device.
  • iOS. Learn about YOLO CoreML models accelerated on Apple's Neural Engine on iPhones and iPads.
  • Android. Explore TFLite acceleration on mobile devices.
• Inference API. Understand how to use the Inference API for running your trained models in the cloud to generate predictions.
  Thanks for the detailed report + screenshots 📱🙏—the missing Export/Download and Favorite buttons in the model details view, plus the “confirm download → nothing happens” flow, is helpful context.
  This is an automated response 🤖, and an Ultralytics engineer will also assist soon.
  If this is a 🐛 Bug Report, please provide screenshots and steps to reproduce your problem to help us get started working on a fix.
  For this specific issue, could you also share a Minimum Reproducible Example (MRE) in the format from the link above, including:
  -Exact steps from a fresh app launch to reproducing the missing buttons (you already listed steps—please confirm if they reproduce 100% of the time)
  -Any additional screenshots/screen recording of the “confirm download” action and what happens afterward 🎥
  -The model type and export format you’re expecting to use (and whether it was ever exported on web)
  -Anything that differentiates this model from others in the same project (e.g., created before/after an app update)
  If this is a ❓ Question, please provide as much information as possible, including dataset, model, environment details etc. so that we might provide the most helpful response.
  We try to respond to all issues as promptly as possible. Thank you for your patience!

[pderrenger]

Thanks for the detailed repro + device/app info — this definitely doesn’t look like expected behavior. On Android, the Ultralytics HUB App should let you select a trained model and download the mobile-ready artifacts, and the model details screen should expose actions like download/export/favorite when the model is compatible and fully synced. (docs.ultralytics.com)

A few quick checks that will help us narrow down whether this is (a) device/OS background restrictions, (b) model-type compatibility, or (c) a model-specific sync/API issue:

1) Confirm the model “task” + architecture

In Ultralytics HUB (web), open the model and tell us:

• Task (detect / segment / classify / pose / obb)
• Architecture (e.g., Ultralytics YOLO11…)

The mobile app experience is best aligned with the core tasks outlined in the app docs, and if the model/task isn’t currently supported in-app, UI actions may be hidden/disabled. (docs.ultralytics.com)

2) Try forcing a mobile export once (workaround test)

Even though web export works, this test helps confirm whether the model can produce the expected mobile artifact:

• In Ultralytics HUB web: Model → Export → TFLite (Android) and wait for it to finish
• Then return to the app and try downloading/using the model again

(For reference, Ultralytics HUB supports exporting models to multiple deployment formats. (docs.ultralytics.com))

3) OxygenOS background/battery settings (common “download does nothing” cause)

On some devices, background work/downloads can be paused aggressively. Please try:

• Settings → Apps → Ultralytics HUB → Battery → set to Unrestricted
• Ensure Background data is allowed
• Re-try the model download while keeping the app in the foreground (screen on) for ~30–60 seconds

4) Sanity check: is this model-specific?

• Do pretrained models download/run in the app for you?
• Do other trained models (different project) show the missing buttons correctly?

If only this one model is affected, it strongly suggests a model metadata/state mismatch being returned to the app.

5) What we need from you to investigate

Please paste:

1. The Model ID (from the Ultralytics HUB web URL, the string after /models/…)
2. The task (detect/segment/classify/pose/obb)
3. Whether the model is Private/Unlisted/Public
4. If possible, an Android logcat snippet captured right after you tap “confirm download” (even 30–50 lines around the event helps)

Also: you mentioned you’re on Ultralytics HUB 0.9.36 — that appears to be the latest Android build currently published. (apkpure.com)

Once you share the Model ID + task, I can route this to the right place internally and we can confirm whether the API is returning an unexpected “actions” set for that model on Android 15.

[MrMati]

Model ID is jt6MucqgEx2AF8sGntnb

It's private as I wasn't able to set project/model to public. It's is a detect YOLO11, other models visible on screenshot are of the same kind. They worked fine, because I triggered the export+download in the Projects tab, before selecting them in the Predict view.

Every functional model exported until this point has a Ultralytics Android export.

But there is only TFLite in web interface.

The export succeeded as you can see. But the Export/Download button in Projects tab is still missing.

Camera view's model selection:

When I select it and accept download, it actually downloads and runs. Silence in logcat during download. So TFLite is enough, I guess.

Also fine in Projects tab:

When I clear data of the app, download/export buttons go missing again.

[pderrenger]

Thanks — the extra screenshots + the fact that it does download and run from the Predict (camera) view is a really helpful signal.

What’s likely happening

• Your export is fine. Since the model downloads and runs in the HUB App after you confirm download, the underlying artifact required for Android (TFLite) is usable. See the Ultralytics HUB App docs for the intended “train in HUB → run on mobile” workflow.
• The missing Export/Download/Favorite buttons in the model details / Projects tab looks like a UI state / sync issue (i.e., the app isn’t correctly recognizing “available actions/exports” for that model on a fresh app state).
• The “Ultralytics Android” export label you’ve seen previously is essentially the app’s “mobile-ready” export option; on web you may only see “TFLite” even though the app can still run it. Exporting to TFLite is the expected path for Android deployment. See the TFLite export guide and the Ultralytics HUB Models docs (export + deploy sections).

Quick workaround (until we fix the UI)

What you already discovered is currently the best workaround:

• Download the model once from the Predict/camera screen, then use it normally.

If you want a “force refresh” attempt that sometimes helps after clearing app data:

1. Open Projects → open the model → stay on the model page ~10–15s
2. Background/foreground the app once (or force close + relaunch)
3. Re-open the model details page and see if actions appear

A couple targeted questions to confirm the exact bug

After you clear app data and sign back in:

1. Do the buttons stay missing for all models, or only for jt6MucqgEx2AF8sGntnb until you download it once from Predict?
2. Once you download it from Predict and it runs, do the buttons then appear consistently for that model until the next clear-data?

What I’ll do next (with your Model ID)

You’ve given everything we need to escalate this internally:

• Model ID: jt6MucqgEx2AF8sGntnb
• Android 15 (OxygenOS 15), OnePlus Nord 4
• HUB App 0.9.36
• Repro involves “first-time use from Predict” + missing actions in model details/Projects after app data reset

If you can add one more data point, it’ll help us pinpoint the API/UI transition:

• The approximate timestamp + timezone of when you open the model details page and see no buttons (e.g., “Dec 18, 2025 10:15 AM CET”), so we can correlate server-side logs.

Appreciate the thorough repro — this is exactly the kind of report that helps us tighten up the HUB App experience.

[MrMati]

1. jt6MucqgEx2AF8sGntnb is consistently the only one with missing buttons after app data reset
2. Yes

The two screenshots in my first post were taken on Dec 13, 2025 8:34 AM GMT

[pderrenger]

Thanks — that confirms it’s not a general app permissions / Android 15 issue, but a model-specific “actions/exports recognized” state bug tied to model jt6MucqgEx2AF8sGntnb.

What your answers tell us

• After an app data reset, only this model loses the Download/Export/Favorite actions in the Projects → Model details view.
• As soon as you “prime” it once via Predict (camera) → confirm download, it downloads + runs, and then the model behaves normally until the next app data reset.
• Your timestamp Dec 13, 2025 8:34 AM GMT is perfect for correlating server-side logs with the app’s initial “fetch model/actions” call.

Given that the model runs fine after the Predict download, the TFLite artifact is valid (and TFLite is the expected Android deployment path; see the YOLO11 → TFLite export guide and the Ultralytics HUB Models export/deploy docs). The gap is that the Projects/model-details UI isn’t being offered actions for this model on a cold start, until a Predict-triggered flow updates local state.

Also, the “Ultralytics Android” label you’ve seen previously is essentially an app-friendly presentation of the mobile export; on the web UI you may only see “TFLite”, even though that’s enough for the Ultralytics HUB App to run the model.

Workaround (until fixed)

Your discovered workaround is currently the best one:

• After reinstall / app data reset: Predict → select model → confirm download (this “primes” the model state for the app).

One last optional check (helps isolate root cause)

If you’re willing, after a fresh app data reset (before priming via Predict), can you try re-exporting TFLite once more for that same model in Ultralytics HUB web (even if it already says succeeded), then immediately open the model details page in-app and see if actions appear without going through Predict?

This will tell us whether it’s:

• purely a client-side caching/state issue, or
• a server-side “actions” payload issue that only corrects after a Predict-triggered request.

Next step on our side

I’m escalating this internally with:

• Model ID: jt6MucqgEx2AF8sGntnb
• Device: OnePlus Nord 4, OxygenOS 15 / Android 15
• App behavior: missing actions after app data reset only for this model, resolved after Predict download
• Timestamp: Dec 13, 2025 08:34 GMT

When a fix lands, we’ll ask you to retest on the latest Ultralytics HUB App build available in the Play Store (since this looks like a very specific UI/state regression).

Appreciate the super crisp repro + timestamps — this is exactly what we need to get it nailed down.

[MrMati]

Re-export doesn't change anything in app ui.

[pderrenger]

Got it — thanks for trying that. If a fresh TFLite re-export from Ultralytics HUB web doesn’t restore the buttons, that strongly suggests this is not an export-job / server-side “missing artifact” problem, but a client-side state bug in the Android app’s Projects → Model details UI for that specific model ID.

What this rules out / confirms

• Rules out: “TFLite export didn’t finish / didn’t exist on the server” as the cause (since re-exporting changes nothing).
• Confirms: The app can still generate/use what it needs via the Predict → download flow, but Projects → Model details is incorrectly deciding “no actions available” on a cold app state only for jt6MucqgEx2AF8sGntnb.

One advanced workaround (optional) that may make it “stick” across app data resets

You mentioned other models show an “Ultralytics Android” export, while web shows only TFLite. Internally, the HUB API supports some “mobile-friendly” export identifiers (e.g. ultralytics_tflite) in addition to standard formats. See the export_fmts_hub() list in the HUB Python reference (it includes ultralytics_tflite).

If you’re comfortable with Python and already have your HUB auth configured locally, you can try triggering that export explicitly:

from ultralytics import hub

hub.export_model(model_id="jt6MucqgEx2AF8sGntnb", format="ultralytics_tflite")

No worries if you’d rather not do this—your current Predict “prime + download” workaround is totally valid.

What I need next (to get this fixed cleanly)

If you can share one more thing, it’ll help our mobile team pinpoint the exact branch that’s hiding the actions:

• After app data reset, when you open Projects → that model → missing buttons, can you grab a logcat while opening that model details page (not during download)? Even ~50–100 lines around the moment the screen loads can help (network response / JSON parsing / feature-flag eval).

If it’s easier, you can also just confirm whether the app shows any “loading/syncing” indicator on that model details page before it renders the empty-actions state.

Next step on our side

I’m flagging this as a model-specific Android UI regression with your confirmed repro + the Dec 13, 2025 08:34 GMT reference time, and we’ll get a fix into an upcoming Ultralytics HUB App release. Once there’s a build out, I’ll ask you to re-test on the newest Play Store version.

Thanks again for staying on it—this is a really high-signal report.

[github-actions]

👋 Hello there! We wanted to give you a friendly reminder that this issue has not had any recent activity and may be closed soon, but don't worry - you can always reopen it if needed. If you still have any questions or concerns, please feel free to let us know how we can help.

For additional resources and information, please see the links below:

• Docs: https://docs.ultralytics.com
• HUB: https://hub.ultralytics.com
• Community: https://community.ultralytics.com

Feel free to inform us of any other issues you discover or feature requests that come to mind in the future. Pull Requests (PRs) are also always welcomed!

Thank you for your contributions to YOLO 🚀 and Vision AI ⭐