DEDUPED Hot AI repos, turned into use cases Get the Monday 5

How to run a private GGUF model on your phone with no cloud with Pocket Strike

Download a small quantized model and chat with it fully on-device.

Authorization notice. Pocket Strike is dual-use. Only use it on software, sites and systems you own or have explicit permission to test.

Reverse engineering. The uncensored presets remove alignment refusals, so treat their output as unfiltered and check it before acting on it.

Who it is forA person who wants to use an AI assistant on an Android phone without sending prompts or data to any cloud service.

The problemCloud chat apps send every prompt off the device. You want answers that stay on your phone.

The outcomeA chat session answered by a local model, with live tokens-per-second shown in the response.

What you need first

  • Flutter SDK >= 3.29.0 and Dart SDK >= 3.12.0
  • Android SDK and NDK with compileSdk 37 and minSdk 26
  • A physical Android device connected over USB/ADB, or an emulator
  • Enough free phone storage and RAM for the GGUF model you choose, as listed in the model table

Steps

  1. Build and install Pocket Strike on your Android device using the Flutter commands in the README, starting with flutter pub get and flutter run -d (add your device ID after -d).
  2. Open the local models area and pick a preset such as Qwen2.5-0.5B-Instruct, which lists about 450 MB RAM footprint in the README table.
  3. Download the model in the app, then tap Load into RAM.
  4. Start a new conversation and send a prompt.
  5. Check the bottom-right of the AI response for the tok/s speed badge.

Prompt for your agent

In the PocketStrike repo, help the user install the app on a connected Android device with flutter run -d <device-id>, then download the Qwen2.5-0.5B-Instruct preset and load it into RAM. Expected output: a chat reply generated locally with a tok/s badge shown. Verify that the reply appears with the model loaded and no cloud provider selected.

Caveat. Model downloads come from HuggingFace and need storage and RAM that match the model size in the README table, so check the footprint before loading a larger model.

Share: X LinkedIn Hacker News Email Copy link

Disclaimer

Deduped lists open-source repositories that belong to other people. We do not own, maintain, audit, or vouch for any of them, and a repo appearing here is not an endorsement or a safety check. The use cases, steps, and prompts are written by AI and may be incomplete, outdated, or just wrong. Where a use case is marked "Seen in the wild", we link to public posts where someone described something similar, but we have not verified those posts and the use case itself is still our AI's own write-up. Use all of this at your own risk. Only run code, commands, or agents on systems and accounts you are authorized to use, and read the code and every command before you run it, especially anything that touches credentials, files, or the network. Nothing here is professional, security, legal, or financial advice. All trademarks and project names belong to their respective owners, and their use here does not imply affiliation. Content is provided as is, without warranties of any kind. Found a mistake, a safety problem, or something you want removed? Email [email protected] and we will look at it promptly. Full disclaimer page