Direct Answer: RAMDeck proves that remote computers miles apart can collaboratively stream 30B AI models at 31 tokens/sec with zero cloud bills. In an uncut back-to-back benchmark, pairing a local 12GB RTX 3060 Mini PC with a remote MacBook Air over standard Wi-Fi and built-in Tailscale generated 31.0 tokens per second on an 18GB model—virtually identical to the 30.0–32.4 tokens/sec measured on all-local Ethernet. The RAMDeck Hardware Hub ($199 • includes 1 Free $29 Zero-Install Node USB) gives developers and privacy-conscious professionals the ability to pool home and office hardware into an encrypted private AI cluster in under 5 minutes with zero technical skills.
From Living Room Cluster to Remote Work Supercomputer
Whether you work from home, a satellite clinic, an accounting practice, or a law office, RAMDeck connects your machines without sending documents to third-party cloud servers:
- Pooled Memory Capacity: A 12GB home desktop and a 16GB office Mac combine into a 28GB pool that runs 30B MoE models smoothly.
- Mobile Road-Warrior Access: The official Apple App Store and Google Play companion apps automatically switch from Home Wi-Fi to encrypted Tailscale when traveling on 5G/LTE.
- Complete Confidentiality: Fully air-gapped from cloud AI providers; zero token metering, zero subscription tiers, and zero data logging.
Explore the full A/B benchmark report at Can You Combine RAM & GPUs Over the Internet? and verify cluster sizing with the Local LLM RAM Calculator.
— The TradeMAV Team
Trademark Notice: TradeMAV™ is a trademark of TradeMAV LLC. All third-party product and company names, including Tailscale, WireGuard, Petals, Ollama, LM Studio, Apple, Mac, Windows, Linux, Qwen, DeepSeek, and Llama, are trademarks or registered trademarks of their respective owners, which are not affiliated with, sponsoring, or endorsing RAMDeck or TradeMAV LLC.
Leave a Reply