arthai-marketplace

Prerequisites

What your machine needs before you pick a stack preset in /arth-setup. This is the short hardware/platform reference — for the full walkthrough (setup questions, task division, keys, feature switches), see the Single-User Guide.

Hardware by preset

Preset RAM GPU Disk Notes
Claude only (default) 8 GB floor not required ~1 GB (toolkit) all inference is remote
Claude + one cloud provider 8 GB floor not required ~1 GB (toolkit) same footprint — inference stays remote
Claude + one local runtime 16 GB minimum, 32 GB recommended Apple Silicon, or NVIDIA with 8–24 GB VRAM on Windows/Linux +50 GB for model weights the only preset with a real local footprint

Running Arth Intelligence and/or arth-memory locally (Docker) adds +2–4 GB RAM on top of whichever preset you picked, and needs no GPU.

The headroom rule, in one line: a local model may claim at most total RAM − 12 GB, and setup only ever offers models that actually fit — see the Single-User Guide §2 for the full rule and how to check free RAM.

Local runtime: RAM-derived default model

Your RAM Default model Alternates
16 GB Qwen-Coder 7B (quantized) —
32 GB+ Qwen3-Coder 30B-A3B (quantized) Devstral, gpt-oss-20b

Every recommendation is gated by the conformance probe before it’s wired into an execution lane — see the Single-User Guide §5.

Platform support

Platform Support
macOS, Apple Silicon native — every preset, every layer
Linux native — every preset, every layer
macOS, Intel cloud presets only — no local-runtime preset
Windows, via WSL2 supported — Ollama/Docker inside the distro (or Ollama Windows-native, reached over localhost)
Windows, native (no WSL2) out of scope