alt.hn

2/14/2026 at 10:39:24 PM

Show HN: Off Grid – Run AI text, image gen, vision offline on your phone

 https://github.com/alichherawalla/off-grid-mobile

by ali_chherawalla

2/15/2026 at 1:45:53 AM

On my Samsung phone it doesn't move the screen up to make room for the keyboard so I can't see what I'm typing.

Really awesome idea though. I want this to work.

by resonious

2/15/2026 at 3:15:32 AM

I can confirm this bug on a Samsung S24 Ultra.

by dotancohen

2/15/2026 at 3:47:42 AM

sorry about that one. I'm taking a look and fixing it right away

by ali_chherawalla

2/15/2026 at 4:43:35 AM

hey, just pushed a fix for it here: https://github.com/alichherawalla/off-grid-mobile/releases/t...

Thanks for spotting and reporting this.

by ali_chherawalla

2/15/2026 at 2:37:13 PM

Great work! I can confirm that the issue is resolved on the Samsung S24 Ultra.

by dotancohen

2/15/2026 at 6:12:56 PM

awesome. thanks!

by ali_chherawalla

2/15/2026 at 12:48:41 AM

Very nice, but I'm gonna wait for the f-droid build.

by bkmeneguello

2/15/2026 at 9:37:08 AM

I've found using Obtanium to be a great tool for tracking github android apps that aren't listed on repos like f-droid

by saratogacx

2/15/2026 at 3:47:20 AM

ok, let me figure that one out.

by ali_chherawalla

2/15/2026 at 3:09:31 AM

Putting the power and the data of the users in the hands of the users themselves! Well done. Getting it setup was easy. Wish the app recognized the keyboard and realized when it was displayed so the bottom menu and chat box weren't hidden under it.

by sangaya

2/15/2026 at 3:38:28 AM

Thank you!

by ali_chherawalla

2/16/2026 at 1:26:00 AM

I gave it a try on Graphene OS: I downloaded models 1st, then blocked Network Accesses. The Apps works fine for text generation, but not for image generation: it needs network access... I did not search where it needs to go.

by Swauzie

2/16/2026 at 3:34:22 AM

it doesn't need internet to generate an image. it needs it to download the model, and to be fair if you've got the zip for the model already you can just import that

by ali_chherawalla

2/16/2026 at 4:16:05 AM

I have download the model... only then I have disabled network access. It loops with an error message needing access to DSP ?

Anyway, I like what you did, as I do not want to train AI with my inputs this is perfect.

by Swauzie

2/16/2026 at 6:06:13 AM

I think thats a bug, I'm guessing you're trying one of the NPU models? If you drop to CPU for now it should hold.

Only phones with qualcomm chips are able to use the NPU. I'm working towards changing that.

but yeah just to be clear there is no internet needed to run any of this. Infact I'm so averse to it, I've not even added analytics for this one. So flying pretty blind here.

by ali_chherawalla

2/15/2026 at 8:34:20 AM

Hasn't here been few of those already eg [1][2]?

1: https://github.com/a-ghorbani/pocketpal-ai

2: https://github.com/shubham0204/SmolChat-Android

by numpad0

2/15/2026 at 8:48:25 AM

i don't think they do image gen etc? You can almost think of Off Grid like an on-device assistant. It'll handle everything like vision, and attachments too. We're still early stages and lots of optimizations to do, but we'll get there.

by ali_chherawalla

2/15/2026 at 3:19:00 AM

Is there something similar, but geared towards a Linux desktop / laptop? I suppose this would be relatively easy to adapt.

by nine_k

2/15/2026 at 10:15:14 AM

If you have enough (V)RAM you can set up Ollama + any of the generation tools here (https://docs.openwebui.com/category/create--edit-images) and hook them into OpenWebUI to get something pretty close to what you'd get on a cloud AI chatbox.

Decently performing models for day to day use cases use at least 6GB VRAM each, though, and even then they're not coming very close to what the cheapest AI websites offer.

by jeroenhd

2/15/2026 at 3:38:49 AM

LM Studio solves for it pretty well I think. It doesn't do image gen etc though

by ali_chherawalla

2/15/2026 at 10:08:55 AM

Looks useful, though something went wrong doing NPU image generation on my phone.

Reminds me a lot of https://github.com/google-ai-edge/gallery which is a proof-of-concept app by Google themselves for their AI libraries. However, your app supports more and larger models without having to manually import anything, which is very useful.

by jeroenhd

2/15/2026 at 10:19:26 AM

i'm adding another feature to recommend models based on h/w. that should take care of this stuff

by ali_chherawalla

2/15/2026 at 1:16:37 AM

The repository is listed as offgrid-mobile everywhere on that page but is off-grid-mobile.

So the lastest releases is at https://github.com/alichherawalla/off-grid-mobile/releases/l...

And the clone would be: git clone https://github.com/alichherawalla/off-grid-mobile.git

by flyingkiwi44

2/15/2026 at 3:38:20 AM

hey, yes. thanks just pushed that fix out

by ali_chherawalla

2/15/2026 at 5:55:59 AM

Looks like the build requirements for SDK and NDK in the build instructions don't match the build.gradle.

-Android SDK (API 34)

-Android NDK r26

vs

compileSdkVersion = 36

targetSdkVersion = 36

ndkVersion = "27.1.12297006"

by flyingkiwi44

2/15/2026 at 6:28:08 AM

yup just updated. Thanks!

by ali_chherawalla

2/15/2026 at 9:01:29 AM

Game-changer.Just downloaded it and wow,this is the future. It’s incredible to have this much power fully offline. No lag, no privacy concerns, just pure utility. I’ve been playing with the vision AI all morning and it hasn't crashed once. 10/10. I have also used other GitHub projects but this feels premium, faster and optimised. Thank you for this!

by ruthlesssavage

2/15/2026 at 9:19:46 AM

thank you, appreciate it!

by ali_chherawalla

2/15/2026 at 1:06:26 AM

I haven't run it, but I looked through the repo. It looks very well thought out, the UI is nice. I appreciate the ethos behind the local/offline design. Cheers.

by derac

2/15/2026 at 3:47:08 AM

thank you!

by ali_chherawalla

2/15/2026 at 10:44:14 AM

I like it, I have a couple minor things (android)

Markdown rendering would be great Letting me download a model, wait for it, then. tell me I can't run it safely in my ram feels like it could've told me at the beginning

Bar that nice! Was it annoying integrating with apples AI stuff?

by willx86

2/15/2026 at 10:49:54 AM

hey! thank you for the feedback. appreciate it. I pushed something recently that actually prevents showing up models in the download screen that don't fit on your phone. Please give it a whirl.

Yeah it wasn't straight forward figuring out the speed + getting it to work for both Android and iOS. Fair bit if complexity. But I'm so happy I got it done.

by ali_chherawalla

2/15/2026 at 10:44:58 AM

I was looking for something like this. Are you open for contributions?

by isawczuk

2/15/2026 at 10:50:32 AM

awesome. The more the merrier. Feel free to get started.

I'm still working on a few things, but let me get a backlog in place and push it so people know what the roadmap looks like

by ali_chherawalla

2/15/2026 at 9:49:15 AM

Are there any nice TTS models? It could be handy to have them as well.

by romperstomper

2/15/2026 at 10:46:22 AM

Check out "FUTO" keyboard Best on device STT I've ever used

by willx86

2/15/2026 at 10:52:26 AM

sure, I'll take a look at it. Thank you for the recommendation

by ali_chherawalla

2/15/2026 at 9:59:58 AM

I don't really do TTS right now, but i allow for STT using whisper. TTS is part of the backlog and I should be adding it soon

by ali_chherawalla

2/15/2026 at 1:23:15 AM

Ok it lists the instruction to build for iOS, but how to sideload?

by vachina

2/15/2026 at 3:04:12 AM

Wonder if you can use GitHub actions to build iOS.

I found a guide for virtual box macOS which failed on intel then another for hyper-V but haven’t tried that one yet.

by instagib

2/15/2026 at 1:12:56 AM

GitHub Releases link is broken.

The dash in "off-grid" is missing.

by snicky

2/15/2026 at 3:46:55 AM

yup, just took a look at that and fixed it. My bad!

by ali_chherawalla

2/15/2026 at 8:56:34 AM

tried this on pixel 9, interesting stuff

by ucguy4u

2/15/2026 at 9:20:11 AM

awesome

by ali_chherawalla

2/15/2026 at 4:00:58 AM

This rules. Godspeed!

by chr15m

2/15/2026 at 4:01:32 AM

wow. thank you

by ali_chherawalla

2/15/2026 at 5:35:28 AM

Nice idea, but isnt this kind of daft?

There are basically no useful models that run on phone hardware.

> Results vary by model size and quantization.

I bet they do.

Look, if you cant run models on your desktop, theres no way in hell they run on your phone.

The problem with all of these self hosting solutions is that the actual models you can run on them aren't any good.

Not like, “chat gpt a year ago” not good.

Like, “its a potato pop pop” no good.

Unsloth has a good guide on running qwen3 (1), and the tldr is basically, its not really good unless you run a big version.

The iphone 17 pro has 12GB of ram.

That is, to be fair, enough to run some small stable diffusion models, but it isnt enough to run run a decent quant of qwen3.

You need about 64 GB for that.

So… i dunno. This feels like a bunch of empty promises; yes, technically it can run some models, but how useful is it actually?

Self hosting needs next gen hardware.

This gen of desktop hardware isnt good enough, even remotely, to compare to server api options.

Running on mobile devices is probably still a way away.

(1) - https://unsloth.ai/docs/models/qwen3-how-to-run-and-fine-tun...

by noodletheworld

2/15/2026 at 6:04:58 AM

The app is basically just a wrapper that makes it super easy to set this up, which I'm very thankful for. I sometimes want to toy with this stuff but the amount of tinkering and gluing things together needed to just get a chat going is always too much for me. The fact that the quality of the AI isn't good is just the models not being quite there yet. If the models get better, this app will be killer.

If there's a similar app for desktop that can set up the stronger models for me, I'd love to hear about it.

by resonious

2/15/2026 at 6:17:12 AM

LM Studio does it well. Along with being a system integrator for SD, and text models I've tried to create a very good chat experience. So theres some sauce over there with Prompt enhancements, Auto detection of images, English Transcription suppor, etc

by ali_chherawalla

2/15/2026 at 10:52:55 AM

> If the models get better, this app will be killer.

Any random thing might happen in the future.

That doesnt have any bearing on how useful this is right now.

All we can do is judge right now how this compares to what it promises.

by noodletheworld

2/15/2026 at 9:11:37 AM

Yeah. The solution if you want to have your own AI is to put a box online or rent cloud inference, and access it over a browser or a phone app.

We have on-prem AI for my microgrid community, but it’s a nascent effort and we can only run <100b models. At least that size is extremely useful for most stuff, and we have a selection of models to choose from on openAI /ollama compatible API endpoints.

by K0balt

2/15/2026 at 6:19:03 AM

I actually think you should give it a spin. IMO you don't need claude level performance for a lot of day to day tasks. Qwen3 8B, or even 4B quantized is actually quite good. Take a look at it. You can offload to the GPU as well so it should really help with speed. Theres a setting for it

by ali_chherawalla

2/15/2026 at 10:51:23 AM

> Qwen3 8B, or even 4B quantized is actually quite good.

No, it’s not.

Trust me, I don't write this from a position of vague hand waving.

Ive tried a lot of self hosted models at a lot of sizes; those small models are not good enough, and do not have a context long enough to be useful for most everyday operations.

by noodletheworld

2/15/2026 at 8:53:06 AM

> The iphone 17 pro has 12GB of ram.

I'm surprised Apple is still cheaping out on RAM on their phones, especially with the effort they've been putting into running AI locally and all of their NPU marketing.

by jeroenhd

2/15/2026 at 9:21:35 AM

with the metal infra its actually quite good. Agreed you can't run really large models, but inference is very fast and TTFT is very low. It's a beautiful experience

by ali_chherawalla

2/15/2026 at 5:38:25 AM

I think if people will people know how accessible it is to run local LLMs on their device they will consider buying devices with more memory that will be able to run better models. Local LLMs in the long run are game changers

by iddan

2/15/2026 at 6:20:13 AM

I agree. I mean mobile devices have only been getting more and more powerful.

by ali_chherawalla

2/15/2026 at 6:17:45 AM

It seems like a good solution for those living under a regime that sensors communication, free information flow, and LLM usage. Especially with a model that contains useful information.

by ImPostingOnHN

2/15/2026 at 5:38:07 PM

[dead]

by altcunn

2/15/2026 at 6:17:56 AM

[dead]

by imposter

2/15/2026 at 4:04:20 AM

[dead]

by wittlesus

2/15/2026 at 4:14:03 AM

This sounds exactly like Claude wrote it. I've noticed Claude saying "genuinely" a lot lately, and the "real killer feature" segue just feels like Claude being asked to review something.

by durhamg

2/15/2026 at 2:10:33 PM

> The fact that you're getting 15-30 tok/s for text gen on phone hardware is wild — that's basically usable for real conversations.

Wild how bad it is compared to, say, Russet for iOS/ipadOS, which runs these same models at 110 tps.

by Terretta

2/15/2026 at 4:12:36 AM

I've added a section for recommended models. So basically you can chose from there.

I'd recommend going for any quantized 1B parameter model. So you can look at llama 3.2 1B, gemma3 1B, qwen3 VL 2B (if you'd like vision)

Appreciate the kind words!

by ali_chherawalla

2/15/2026 at 4:11:35 AM

> that's basically usable for real conversations.

That's using the word "real" very loosely.

by add-sub-mul-div