Termux AI

While I was using the Termux app to monitor my bazaar shop while flying I come across an article about running a miniture AI in mobile devices using the Termux app.

I read the article dating back to 2024, I followed the tutorial on how to run it and fail several times but eventually I reach 98% progress --- It was in deep into the nigth, I was determine to make it run before I go to sleep what I didnt expect is that when I type the command to run it --- It froze my screen and suddenly my phone started to heat up I was panicking inside on what to do luckily my mobile device was low battery, I was gonna put my phone in the refigirator like a goddamn supercomputer about to lit up, scary experience.

Then suddenly after the initializing of the Llama-3.2-1B-Instruct-Q3_K_L.gguf interface in Termux app, I was able to take control again.

I didnt dare ask the llama.ai any question, I dont want my phone to burn in my hand but the Curiosity killed the cat.

01/03/2026 📟 almost like a pager in middle east 🧨

[image: i.ibb.co]

Ji

Bit of confusion but i suppose its informitize non the less

I really don't know what to make of all of that. Sounds like a very strange experience

Probably a coincidental glitch. Who knows?

I suggest you use Operit AI instead, it's the most powerful personal android software available right now. It was developed by AAswordman on Github and is still being actively updated with new patches and fixes, it's also open-source. You can either use local models, or cloud-based ones from several providers. I'd recommend you stay away from using local models, their performance aren't worth it when you will be able to accomplish tasks and workflows far faster with a cloud model, the comparison is night and day between frontier state of the art cloud models and smaller local models.

If you can't pay for any cloud providers, use the Nvidia NIM API, which is free, only a phone number is required to create an account with them. You will need an API Key (it's free) from their API, and then you have to switch your provider in the model settings to the Nvidia API, afterward—copy and paste the Nvidia NIM's API key into the Operit AI model API key box. 

The software gives all LLMs agentic capabilities, even if a model doesn't natively support tool use.