What can your computer run?

Answer four quick questions and we'll email you a free, personalized setup report, written by a human, in plain English, for your exact situation.

Live right now

See what your computer can run

This list updates itself from the open-source community and picks models that fit your device. No hardcoded guesses.

Thinking about a GPU?

What graphics cards actually buy you

Every platform gets you there a different way. Pick a vendor tab to see the lineup and the jobs each card handles.

Mac (Apple Silicon)

One shared memory pool for processor and graphics: quiet, efficient, always-on friendly. Apple-optimized apps unlock bigger models than you'd expect from a laptop.

NVIDIA

The speed-per-dollar king. CUDA is the best-supported AI stack, so everything just works. From a used 3060 to the 5090, this is the default upgrade path.

AMD

Strong value. The RX 7900 XTX rivals the big NVIDIAs for less money, and AMD's software support keeps improving month by month.

Intel

The budget wildcard. Arc cards like the B580 put 12 GB of VRAM in reach for the price of a nice dinner out. Fine for everyday models.

AI PCs

Appliance simplicity: DGX Spark, Strix Halo minis, Mac Studio. One big memory pool, zero tinkering. More below.

CardUsable memoryThe jobs it handles
RTX 306012 GB VRAMEveryday chat, summaries, homework help, social drafts
RTX 4060 Ti16 GB VRAMAbove + real coding help, document analysis
RTX 309024 GB VRAMAbove + deep research, content production at speed
RTX 409024 GB VRAMAbove + image generation, video-gen experiments
RTX 509032 GB VRAMEverything above, faster + serving the whole house
RTX 6000 Ada48 GB VRAM70B models for a home office or small business

Going bigger? Get the Home Lab Buyer's Guide

AI PCs vs. multi-GPU builds, what 512 GB actually buys you, real prices, and the mistakes to avoid. Free, straight to your inbox.

Read the buyer's guide
First week expectations

The honest fine print

A few things everyone notices in week one, none of them are problems.

The first load takes a moment

The first time you run a model it loads into memory, like an app opening. After that it's quick. Bigger models load slower; that's normal, not broken.

Your fans might spin up

AI is a workout for your processor. A warm laptop and spinning fans while it answers = it's working hard for you. It calms down between answers.

Close heavy apps for big models

Models live in RAM. If things feel sluggish, close the 47 browser tabs first. 8 GB of RAM comfortably runs everyday models; 16 GB runs the good stuff.

It doesn't know today's news

Local models know what they knew when they were made and don't browse the web by default. For current events, that's the one thing you'll still Google.

The next step

AI agents: from chatting to doing

Chat answers questions. Agents do the work: they watch, decide, and act on your hardware, under your rules.

What agents actually do

Watch an inbox, folder, or feed and handle the routine work automatically: filing, drafting, scheduling, summarizing, running code, chaining tools together. Like a hire who never sleeps and never phones it in.

The popular ones

OpenClaw, Hermes-class agents, and many others, each better at different jobs. The right pick depends on your use case and your industry, which is exactly the kind of thing we map for people.

We build them

Our team and our community build agents for households and businesses, from "watch this folder" helpers to full workflows that run your back office. Locally, privately, on your terms.

Want an agent working for you?

Tell us the task that eats your week. We'll tell you honestly what it takes to automate it, and whether it's worth automating at all.

Talk to us about agents

Get your free setup report.

Four questions, one personalized report, zero obligation.