# Running LLM on Android

**URL:** <https://discuss.huggingface.co/t/running-llm-on-android/144838>\
**Category:** Beginners\
**Created:** [March 9, 2025, 3:31am UTC](https://discuss.huggingface.co/t/running-llm-on-android/144838 "2025-03-09T03:31:08Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![Aljames](https://sea2.discourse-cdn.com/hellohellohello/user_avatar/discuss.huggingface.co/aljames/32/42851_2.png) [@Aljames](https://discuss.huggingface.co/u/Aljames)\
**Post date:** [March 9, 2025, 3:31am UTC](https://discuss.huggingface.co/t/running-llm-on-android/144838/1 "2025-03-09T03:31:08Z")

</div>

**Newbie hobby project**  
running ollama and other llms on a android phone.  
While learning AI along the way.

Used Termux to install proot-Ubuntu and Ollama using ai generated code.

I stopped at loading models while learning models for use cases on public platforms.

If anyone has done similar, what models, sizes and use cases run well on a 12gb phone with 512ssd.

Which mini models could run basic rewriting, grammer, dataset chat, image tagging, digital garden, static sites flow. Etc.  
I’ve tried pocketai app but that’s not great.

Suggestions appriciated

Got a couple of spare 8gb phones that I imagine being better than Raspberry Pi at self hosting.

---

<div class="post-metadata">

**Author:** ![John6666](https://sea2.discourse-cdn.com/hellohellohello/user_avatar/discuss.huggingface.co/john6666/32/27664_2.png) [@John6666](https://discuss.huggingface.co/u/John6666)\
**Post date:** [March 9, 2025, 6:50am UTC](https://discuss.huggingface.co/t/running-llm-on-android/144838/2 "2025-03-09T06:50:52Z")

</div>

When it comes to running on an edge device, I think it’s safest to use a model that is at most 7B in size after quantization, and preferably 3B or less.

> **[Running Llama 3.2 on Android: A Step-by-Step Guide Using Ollama](https://dev.to/koolkamalkishor/running-llama-32-on-android-a-step-by-step-guide-using-ollama-54ig)**
>
> Llama 3.2 was recently introduced at Meta’s Developer Conference, showcasing impressive multimodal...

---

<div class="post-metadata">

**Author:** ![Void2377](https://sea2.discourse-cdn.com/hellohellohello/user_avatar/discuss.huggingface.co/void2377/32/55540_2.png) [@Void2377](https://discuss.huggingface.co/u/Void2377)\
**Post date:** [November 8, 2025, 9:21am UTC](https://discuss.huggingface.co/t/running-llm-on-android/144838/3 "2025-11-08T09:21:24Z")

</div>

Check out **ToolNeuron v5** — an offline AI assistant for Android.

- Run your own GGUF models

- Use plugins like web search, email, scraping

- Text-to-speech & speech-to-text

- Fully offline & private  
Try it here: [Release Tool-Neuron-BETA-5.0 Stable Build · Siddhesh2377/ToolNeuron · GitHub](https://github.com/Siddhesh2377/ToolNeuron/releases/tag/Beta-5.0)