THE PHONES IN YOUR DRAWER JUST BECAME ONE BRAIN
One GGUF model, split across phones you already own. Your laptop chats with it like any other AI.
Your strongest Android leads β it loads the model and runs the chat.
The rest join by QR β another Android, an iPhone, an iPad.
Your PC uses one link β http://192.168.x.y:8080/v1.
Free and MIT, nothing new to buy β github.com/orailnoor/cross-platform-llm-client
Brains that split well β pick by small active slice, not by size:
- 26B-A4B Β· 10 GB β the four-phone demo brain
- 30B-A3B Β· 10 GB β the coding-flavoured one
- 16B-A3B Β· 6.5 GB β for the smaller phones in the pile
π Plug the cluster into apps you already use
- Open WebUI β add a connection with the
/v1URL and the appβsaichat_β¦key - Continue for VS Code β coding with the brain living on your phones
- AnythingLLM β chat with your own files; base path includes
/v1 - ChatBox β a small desktop chat window
- LobeChat β set
OPENAI_PROXY_URLto the/v1URL - Obsidian Text Generator β ask your notes on the private brain
π§© The trick that makes it click
Small active slice wins β a 26B brain wakes about 4B per word, so each phone reads a sliver: the demo brain Β· the coding one Β· the lighter one
whichquant checks the fit from the modelβs header before you download
Pooled runs the same pooling in browser tabs β no install, iPhones join as equals
Tailscale on the leader and laptop reaches the cluster from anywhere
The app also draws images offline and reads photos and PDFs with a vision brain
The makerβs walkthrough shows the QR pairing end to end
Your drawer just got a new job. ![]()

!