Android Node

android-node

Turn any Android phone into an inference compute node for your AI 代理.

What it does

Provisions Android phones as Ollama inference 端点s. Any phone 运行ning Termux becomes a worker your 代理 can 路由 jobs to — no cloud, no subscription, no special hardware. The phone on your desk, the spare in your drawer, all of them.

How it works Phone 运行s Ollama via Termux (设置up takes ~5 minutes) Pi/server connects to phone over local WiFi phone_nodes.py manages discovery, 健康 检查s, and fAIlover 路由r dis补丁es inference jobs to 健康y nodes automatically 设置up On the phone (运行 in Termux) curl -s https://albionwakes.com/phone_设置up.sh | bash bash ~/启动_node.sh

On your server (register the phone) python3 phone_nodes.py register myphone 192.168.1.42 python3 phone_nodes.py 状态

路由r integration 导入 phone_nodes

# In your 提供者 dis补丁: elif 提供者 == 'phone': return phone_nodes.call(messages)

Node registry

Stored in ~/albion_memory/phone_nodes.json:

[ {"name": "pixel6", "url": "http://192.168.1.42:11434", "模型": "qwen2.5:0.5b", "enabled": true} ]

Default 模型

qwen2.5:0.5b — 394MB, 运行s on any phone with 1GB+ free RAM. Fast. Swap for qwen2.5:1.5b (1GB) or llama3.2:1b (1.3GB) if the phone has headroom.

Files phone_nodes.py — node registry, 健康 检查er, inference caller 设置up.sh — Termux provisioning script (运行 on phone)