System Requirements

Preships runs on macOS, Linux, and Windows. Deterministic checks work on any machine with Node 20+. No GPU or model runtime needed for those.

Minimum Requirements (Deterministic Only)

Component	Requirement
Node.js	v20 or later
OS	macOS 13+, Ubuntu 20.04+, Windows 10+
RAM	4 GB minimum
Disk	~500 MB (Playwright browsers + CLI)

Local Model Requirements (Ollama)

For AI-powered checks, Preships uses Ollama for local inference. Here's what each model tier needs:

Model	Params	Min RAM	Best For
`qwen2.5-coder:3b`	3B	4 GB	Light reasoning, fast checks
`qwen2.5-coder:7b`	7B	8 GB	Default — good balance of speed and quality
`llama3.2:11b-vision`	11B	12 GB	Visual checks — screenshot analysis
`qwen2.5-coder:14b`	14B	16 GB	Complex interaction flow reasoning
`qwen2.5-coder:32b`	32B	24 GB	Maximum local quality

Apple Silicon Macs share system RAM with the GPU, so a 16 GB M-series MacBook can comfortably run 7B–14B models. On discrete GPU systems, VRAM is what matters — a 12 GB NVIDIA card handles up to 11B well.

Check Your System

preships info

Detects your hardware, checks for Ollama and Playwright, and recommends which model to use based on available RAM.

Cloud-Only Mode (No Local Model)

preships login --api-key psk_your_key
preships config set provider cloud

Deterministic checks still run locally. Only reasoning/visual checks route through the cloud.