Home About Features Help Pricing ⬇ Download
Download › macOS
macOS

ComIO.Studio for macOS

Apple Silicon (M1 / M2 / M3)

Coming soon
macOS version is not available yet — we will announce the release in our newsletter.

System Requirements

OS macOS 13 (Ventura) or newer
Architecture Apple Silicon only — M1, M2, M3 (Intel Macs not supported)
Unified memory 8 GB minimum (terminal only) — 24–32 GB recommended if you want local AI
Disk space 300 MB free
USB USB port for serial communication — a USB-C to USB-A adapter may be needed

Local AI — Unified Memory optional

The serial terminal, macros, triggers, scripting, and programmer work without any extra hardware. Local AI models powerful enough for ComIO.Studio (gpt-oss 20B, Gemma 4 26B / 31B class) require substantial unified memory. Smaller models under ~13B parameters are too weak for reliable assistance — we do not recommend them.

How Apple Silicon memory works for AI: Unlike PCs where CPU and GPU have separate memory pools, the M-series chip shares a single pool of unified memory between CPU and GPU. The AI model loads directly into the memory your GPU uses — the more you have, the larger the model that fits.
16 GB unified memory Bare minimum — fits gpt-oss 20B at low quantization (Q3-Q4), tight context window
24 GB unified memory Comfortable for 20B-class models (Q4-Q5), workable Gemma 4 26B
32 GB unified memory Recommended — Gemma 4 31B comfortably (Q5-Q6), larger context, room for OS
64 GB+ unified memory High-end (M2/M3 Max/Ultra) — 70B-class models, very large context
Acceleration Apple Metal — automatically used, no configuration needed

Base 8 GB Macs cannot run models powerful enough for ComIO.Studio assistant. For day-to-day AI-assisted serial debugging, target 24–32 GB unified memory.