One free script installs a complete AI stack on any Mac or Windows laptop in about 20 minutes. Works offline. Nothing leaves your device.
curl -s https://gotlaptopparts.com/ai-setup/mac.sh -o /tmp/ai-setup-mac.sh && bash /tmp/ai-setup-mac.shirm https://gotlaptopparts.com/ai-setup/windows.ps1 -OutFile "$env:TEMP\ai-setup.ps1"; powershell -File "$env:TEMP\ai-setup.ps1"Everything runs privately on your laptop.
Everything runs on your laptop. Your prompts, documents, and conversations never leave your device. No company — including GotLaptopParts — can see what you discuss. The script is fully open source. Read every line before running.
Four steps. Most of it runs while you walk away.
All free. All open source. No accounts required.
The script detects your RAM and installs the best model automatically.
| Your RAM | Model installed | Best for |
|---|---|---|
| 48GB+ | Llama 3.1 70B | Maximum quality |
| 24–47GB | Qwen3 32B | Near-frontier quality |
| 12–23GB | Llama 3.1 8B | Full quality, everyday use |
| 6–11GB | Phi-4 Mini | Fast, lightweight chat |
After setup, switch to any of 200+ models inside LM Studio → Discover. No command line needed.
Run our free setup script — one command in Terminal (Mac) or PowerShell (Windows). It detects your hardware, installs Ollama, LM Studio, Jan.ai, and AnythingLLM, downloads the best AI model for your system, and has everything working in about 20 minutes. No account required. No subscription.
For most everyday tasks — writing, summarizing, answering questions, document review — local AI models in 2026 are comparable to ChatGPT. On hardware with 32GB+ RAM or a dedicated GPU, models like Qwen3 32B or Llama 3.1 70B produce near-frontier quality. The main difference: local AI never has rate limits, never goes offline, and never sees your data.
Yes. Once the model is downloaded during setup, your AI works completely offline — on planes, in areas with no WiFi, anywhere. Nothing requires an internet connection after the initial setup.
The script is fully open source at github.com/gotlaptopparts/free-local-ai — read every line before running. It installs software only via Homebrew (Mac) and winget (Windows), both verified signed package managers. It never collects data, never connects to GotLaptopParts servers, and never installs tracking or telemetry of any kind.
Local AI can replace ChatGPT Plus, Perplexity Pro, GitHub Copilot, Claude Pro, and Grammarly Premium. The tools installed — LM Studio for chat, Page Assist for browser AI, Continue.dev for coding, and AnythingLLM for document chat — cover the same use cases as all of these paid services. Your actual savings depend on which subscriptions you currently pay for.
Minimum 6GB RAM for basic AI (Phi-4 Mini). 12GB for full quality (Llama 3.1 8B). 24GB for near-frontier quality (Qwen3 32B). 48GB+ for maximum capability (Llama 3.1 70B). Apple Silicon Macs are especially efficient — a MacBook Pro M2 with 16GB runs Llama 3.1 8B at 30–80 tokens per second.
Yes — Apple Silicon is the best hardware for local AI. Metal GPU acceleration is automatic. The script detects your chip and configures everything optimally. A MacBook Pro M3 with 16GB RAM runs Llama 3.1 8B smoothly for all everyday tasks.
Yes. Open LM Studio, click Discover, and browse 200+ free models. Download any with one click — no command line needed. Different models excel at different tasks: Qwen3-Coder for programming, DeepSeek R1 for reasoning, Gemma 4 for image understanding, Phi-4 Mini for fast lightweight chat.
We sell refurbished laptops with everything configured, model pre-downloaded, and verified working before shipping. Turn it on — AI works immediately.
Shop AI-Ready Laptops →Something not working? Call us or visit our help page.