VTM Studio (Version 2.2) · Desktop Builds & Hardware Guide

Desktop Builds &
System Requirements.

Review target operating system packages, minimum vs. recommended hardware specifications, and single-board computer tuning guidelines for VTM Studio.

APPLE MAC MACOS 10.15+

macOS Installer (.dmg)

Intel Core i5/i7 (macOS Catalina 10.15+) & Apple Silicon (Pending Verification)

Standalone macOS application bundle (VTM Studio.app) with integrated hardware serial tools, audio and image utilities, and offline synthesized voice units.

PACKAGE: VTM-Studio-2.2-macOS.dmg
UPDATES: Signed In-App Maintenance Updates
LICENSE: Good for 2 Devices per Key
Join Priority Waitlist →
SINGLE-BOARD COMPUTERS DEBIAN / UBUNTU ARM64

Raspberry Pi, UNO Q & ARM64 SBC (.deb)

Raspberry Pi 4B/5, Arduino UNO Q & 64-Bit Debian Linux SBCs

Native 64-bit Debian package tested on the Raspberry Pi 4B (4GB) and Arduino UNO Q (4GB), and compatible with other 64-bit ARM64 Debian/Ubuntu single-board computers. Includes RAM disk-cache tuning to reduce MicroSD iowait.

PACKAGE: vtm-studio_2.2_arm64.deb
ARCHITECTURE: 64-Bit ARM64 Linux
UART BRIDGE: 115200 8N1 Serial Sync
Join Priority Waitlist →
WINDOWS PC WINDOWS 10 & 11

Windows 64-bit (.exe)

Windows 10 & Windows 11 (x86_64) · Desktop Installer

Planned 64-bit Windows desktop installer with integrated hardware serial tools and automatic COM port (COM3+) CH340/CP2102 detection.

TARGET: VTM-Studio-2.2-Setup.exe
ARCHITECTURE: 64-Bit x86_64 Windows
STATUS: Pending Physical Verification
Join Priority Waitlist →
● SIGNED IN-APP MAINTENANCE UPDATES

Updates Within the Version You Purchase

Your purchase includes free maintenance and patch updates within the version you purchase (it does not include free upgrades to future major versions). When online, the desktop app checks for signed patch updates in the background:

  • 1. Verifies your active device activation (Device 1 or Device 2).
  • 2. Verifies the update package's cryptographic signature and file integrity before unpacking.
  • 3. Reveals the top-bar [ Restart to Update ] pill and applies the update cleanly on restart.
● LICENSE GOOD FOR 2 DEVICES

Install on Up to 2 Machines You Own

Every VTM Studio license key is good for 2 devices that you own (for example, your primary desktop or laptop plus a secondary computer or single-board computer).

Simply keep the installer from your purchase email and activate VTM Studio on your second device using your same License Key.

03 // MINIMUM VS. RECOMMENDED HARDWARE

System Requirements & Hardware Expectations

Please review the Recommended Hardware column before purchasing. While VTM Studio installs and runs on minimum-spec hardware (including Raspberry Pi 4B, Arduino UNO Q, and older Intel Macs), running the full workstation with all packs loaded will run noticeably slower on single-board computers than on a recommended desktop or laptop. Edge AI Local LLMs are a separate GPU-dependent workload and run very slowly on SBCs and Macs without strong GPU acceleration—on those machines, using Cloud AI (Google Gemini or Groq) is strongly recommended.

Platform Tier Minimum Hardware Recommended Hardware Tested Hardware
macOS (Apple Silicon) Apple M1, 4 GB RAM, macOS 12+ Apple M1–M4, 8–16 GB RAM, macOS 14/15+ Pending Physical Verification
macOS (Intel Modern & Legacy) Intel Core i5/i7, 4 GB RAM, macOS 10.15 (Catalina) Intel Core i7 Quad-Core, 8–16 GB RAM, SATA/NVMe SSD 2017 MacBook Pro (Intel) & 15-inch MacBook Pro (Core i7, 16GB RAM, macOS Catalina)
Embedded SBCs (Linux ARM64) Quad-Core ARM Cortex-A72, 4 GB RAM, Debian 64-bit Quad-Core ARM, 4–8 GB RAM, USB 3.0 SSD Boot Raspberry Pi 4B (4GB) & Arduino UNO Q (4GB RAM / 32GB eMMC, Qualcomm 1.8V MPU + STM32 3.3V MCU)
Windows (Planned Target) Intel Core i5 / Ryzen 64-bit, 4 GB RAM, Windows 10 Intel Core i7 / Ryzen 7, 8–16 GB RAM, Windows 11 Pending Physical Verification
Local Edge AI & Hugging Face LLMs
(Optional — Third-Party Models)
 2–4 GB RAM (SBC) or 4 GB GPU VRAM / 16 GB System RAM
(for 1B–3B quantized GGUF models)
 12 GB+ Dedicated GPU VRAM (NVIDIA CUDA / AMD / Intel Arc) or 16 GB+ Unified Memory
(for 7B–12B+ models without CPU offload)
• 3–8 tok/s (1B Gemma on Arduino UNO Q 2GB)
• 10–15 tok/s (1B Gemma on 2017 MacBook Pro 16GB)
• ~2 tok/s (3B Phi-2 Q8_0 [3.58GB] on Win 10, 4GB GPU, 24GB RAM, SSD)
• ~3.7 tok/s (iGPU) vs. ~32 tok/s (12GB RTX 3060) (12B Gemma)
Note: Provided solely to give you a realistic expectation. Speeds vary depending on model size, quantization (Q4 vs. Q8), context settings, and hardware variables. VTM Studio does not sell third-party LLMs.
RASPBERRY PI, ARDUINO UNO Q & ARM64 SBCS // MICROSD IOWAIT FIX

Eliminate MicroSD Random-Write Lag on SBCs

Standard Class 10 / A1 microSD cards bottleneck at 100–400 IOPS during random 4KB writes, which can cause Linux kernel iowait cursor stalls on single-board computers. For smoother SBC performance, boot from a USB 3.0 SSD and keep temporary disk cache in RAM (tmpfs / /dev/shm).

HARDWARE TROUBLESHOOTING // USB SERIAL CHECKLIST

USB UART Serial Connection Checklist

If your microcontroller port does not appear in Code Mode or Science Mode:

  • • Verify your USB cable carries data lines (D+/D-) and is not a charge-only lead.
  • • Confirm USB-to-UART bridge drivers (CH340, CP2102, or FTDI) are installed on your host OS.
  • • Match your sketch baud rate (115200 baud) with the active serial monitor or telemetry panel.