News for hackers. Stuff that matters.

Show HN: Qwen 3.5 running on a $300 Android phone – on-device, open source

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

Qwen 3.5 Small dropped two days ago. I had it running on a mid-tier Android phone within hours.

It's great seeing the on-device AI community light up around this release. Off Grid brings it to Android: phones with 6GB RAM in the $200-300 range, ~8 tok/sec on the 2B model. Fully offline.

Text generation, vision AI, image gen, voice transcription, tool calling, document analysis — all on-device, nothing uploaded, ever. Works in airplane mode.

780+ GitHub stars. ~2,000 downloads across Android and iOS. Early days.

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

Play Store: https://play.google.com/store/apps/details?id=ai.offgridmobi...

App Store: https://apps.apple.com/us/app/off-grid-local-ai/id6759299882

( 10 comments | 3 points | on HN )
Re:Show HN: Qwen 3.5 running on a $300 Android phone – on-de... by alefiyakachwala on Tuesday March 03, 2026 @03:32PM (#47238544)
Can you share some technical details? How did you do it? What’s under the good?
Re:Show HN: Qwen 3.5 running on a $300 Android phone – on-de... by ali_chherawalla on Tuesday March 03, 2026 @03:34PM (#47238566)
ofcourse ofcourse,

I've documented everything here: https://github.com/alichherawalla/off-grid-mobile-ai/blob/ma...

llama.cpp compiled as a native Android library via the NDK, linked into React Native through a custom JSI bridge. GGUF models loaded straight into memory. On Snapdragon devices we use QNN (Qualcomm Neural Network) for hardware acceleration. OpenCL GPU fallback on everything else. CPU-only as a last resort.

Image gen is Stable Diffusion running on the NPU where available. Vision uses SmolVLM and Qwen3-VL. Voice is on-device Whisper.

The model browser filters by your device's RAM so you never download something your phone can't run. The whole thing is MIT licensed - happy to answer anything about the architecture.

Re:Show HN: Qwen 3.5 running on a $300 Android phone – on-de... by mwze on Tuesday March 03, 2026 @09:02PM (#47242050)
Any roadmap to add Mediatek NPU support?
Re:Show HN: Qwen 3.5 running on a $300 Android phone – on-de... by ali_chherawalla on Tuesday March 03, 2026 @09:25PM (#47242213)
I'm working on that we speak. Shouldn't not be that difficult of a lift and should be able to do that tonight or in the next couple of nights
Re:Show HN: Qwen 3.5 running on a $300 Android phone – on-de... by mwze on Wednesday March 04, 2026 @08:41PM (#47256400)
Happy to test, I have poco X6 pro, 12gb ram model
Re:Show HN: Qwen 3.5 running on a $300 Android phone – on-de... by ali_chherawalla on Wednesday March 04, 2026 @09:45PM (#47256865)
awesome. I'll let you know once thats in
Re:Show HN: Qwen 3.5 running on a $300 Android phone – on-de... by ruthlesssavage on Tuesday March 03, 2026 @07:34PM (#47241335)
You have eliminated the problem of latency and having flagship phones. Amazing, my old android has AI now. Gave you star on github. Ciao.
Re:Show HN: Qwen 3.5 running on a $300 Android phone – on-de... by ali_chherawalla on Tuesday March 03, 2026 @09:25PM (#47242217)
lol thanks buddy!
Re:Show HN: Qwen 3.5 running on a $300 Android phone – on-de... by ucguy4u on Tuesday March 03, 2026 @10:14PM (#47242561)
The current copy function copies the entire message.

Could it please be improved to allow selection and copying of only the desired text?

Re:Show HN: Qwen 3.5 running on a $300 Android phone – on-de... by ali_chherawalla on Wednesday March 04, 2026 @07:35AM (#47246593)
fair point, let me know work on that. That's a small lift.