Direct Answer: Yes, you can combine the RAM of your iPhone, iPad, or Android phone with your Mac, Windows PC, or Linux computer over local Wi-Fi to run 13B, 27B, 30B, and 120B Large Language Models (LLMs). While standalone phone apps (like PocketPal AI or Locally AI) are capped by single-phone memory limits to small 1B–4B models—and DIY frameworks (like Exo or manual RPC) require Linux terminal emulators (Termux) on Android or sideloaded builds on iOS—RAMDeck ($159 Hardware Hub + Free $29 Zero-Install Node USB) is the only system with official 1-click apps on both the Apple App Store (iPhone & iPad) and Google Play Store (Android).
28 Tokens/Sec Over Wi-Fi: Why Adding Phones Doesn’t Have to Mean 1–2 Tokens/Sec
A common misconception on forums is that pooling smartphones over Wi-Fi always drops generation speed to 1 to 2 tokens per second. In our verified lab benchmarks, an 18 GB Qwen 30B Mixture-of-Experts (MoE) model that exceeded the memory of a 16 GB M3 MacBook was pooled over local Wi-Fi with an iPhone 16 (contributing 2.0 GB) and an iPhone XS (contributing 1.0 GB) using the official RAMDeck Apple App Store app:
- Load Time Over Wi-Fi: Under 30 seconds
- Generation Speed: 28 tokens per second (first-token turnaround under 5 seconds — watch uncut video)
- Live Worker Pause & Resumption: Unplugging or pausing the iPhone mid-session gracefully preserves cluster state (watch live pause/resume demo).
Combining Android + iPhone + PC/Mac with Zero Termux
We also pooled a 12 GB Acer laptop, an RTX 3060 Mini PC, a Mac mini, and a Samsung Galaxy S24+ Android phone via the official Google Play app to run a ~16 GB Qwen 3.8 27B Dense model in RAM at 1.92 tok/s (watch the 110,000+ views video) and a 13B Dense model at 11–14 tok/s (watch demo).
Read the complete iPhone and Android hardware tables at Can You Combine Phone RAM (iPhone & Android) to Run a Bigger Local LLM?, inspect our public core engine at github.com/trademav/ramdeck-core-public, and claim your $159 RAMDeck Hub + Free $29 Zero-Install Node USB on Indiegogo.
— The TradeMAV Team
Leave a Reply