BestLLMfor Your hardware. Your LLM. Your call.
The Local Copilot Kit APIOpen data Find my LLM
Tour 01 · Getting started

The beginner — “I just want to try it, no fuss.”

You've heard about local LLMs and you want to see how they run on your own machine. No code, no complicated command line — by the end you'll be chatting with your first model, offline, for free.

Total duration
~25 min
Skill level
None
Cost
Free
Machine type
Recent PC or Mac (8 GB+ RAM)
Our promise

By the end of this tour, Ollama will be running quietly in the background, you'll have picked and downloaded your first model, and you'll have a clean web interface to chat with it — all without a single line of code to understand.

Who this tour is for

We'd rather tell you upfront than let you lose half an hour for nothing.

✓ This is for you if
  • You barely know what an LLM is, but the topic intrigues you.
  • You want to try it without handing your questions to a cloud service.
  • You're looking for the shortest path between “I've heard about it” and “it's running on my machine.”
  • You have a PC or Mac less than 5 years old with at least 8 GB of RAM.
✕ Look elsewhere if
  • ·You want a local coding copilot in your editor: see the developer tour.
  • ·You need to query your own documents privately: see the confidential-work tour.
  • ·Your machine is from 2015 with 4 GB of RAM: it'll run, but it'll crawl.

The path in 6 steps

Each step links to a real, detailed guide. The order is deliberate — don't skip steps the first time through.

Total · ~25 min
  1. 1
    Step 01 · 2 min

    Find out what your machine can run

    Answer a few quick questions about your GPU, RAM and OS and get a shortlist of models sized correctly for your hardware — no guessing.

    You know which model sizes and quantizations will actually run well on your machine.
    Read the guide →
  2. 2
    Step 02 · 5 min

    Understand what “local” really means

    The foundational primer: what a local LLM is, why people bother, and what to realistically expect versus ChatGPT.

    You understand the trade-offs before installing anything.
    Read the guide →
  3. 3
    Step 03 · 3 min

    Install Ollama

    The fastest way to get an inference engine running in the background — download, double-click, done.

    Ollama is installed and listening on localhost:11434.
    Read the guide →
  4. 4
    Step 04 · 3 min

    Pick your first model

    A ranked, up-to-date shortlist of the best free, open-weight models to start with, so you're not guessing which one to download.

    You've picked a model and can pull it with a single ollama run command.
    Read the guide →
  5. 5
    Step 05 · 8 min

    Get a ChatGPT-like interface

    The terminal is austere. Open WebUI gives you a polished chat interface with history, markdown and file uploads — one Docker command.

    You're chatting at localhost:3000 with a familiar, browser-based interface.
    Read the guide →
  6. 6
    Step 06 · 4 min

    See how it really compares to ChatGPT

    Honest expectations: where a local 7–8B model holds up against ChatGPT, and where the cloud still wins.

    You know exactly when to reach for local and when to still open a browser tab.
    Read the guide →

The recommended models

Three picks for this tour, lightest to most capable. Click through for the full spec sheet — VRAM, context, benchmarks.

Frequently asked

The real questions readers send in. If yours is missing, reach out via Contribute below.

Does it use a lot of electricity?

During a conversation: yes, your GPU runs close to full tilt (200–400 W depending on the model). Idle between questions: almost nothing. For occasional use it's negligible next to a gaming session.

Are my conversations really private?

Yes, entirely. Once Ollama and a model are installed, you can disconnect from the internet and it keeps working. Nothing leaves your machine — no telemetry, no account, no analytics.

Is it really free?

Yes. Ollama is open source (MIT), and the models above are open-weights with permissive licenses. No subscription, no quota, no “pro” tier — you only pay for electricity.

Why is it worse than ChatGPT?

A 7–8B model has a few billion parameters; GPT-4-class models likely have well over a trillion. For something that fits in 5–6 GB of VRAM, the gap is smaller than you'd think — and a local 70B model closes most of it.

What if I want to try a different model afterward?

ollama run llama3.2, ollama run qwen2.5, ollama run gemma2 — the full library is at ollama.com/library. Each model downloads once, then launches instantly.

Once this tour is done
Next tour

The developer — “Local copilot, 100% offline, zero cloud latency.”

A local coding model wired into your editor. Autocomplete, chat, whole-file refactors — without a single line of your code leaving your machine for Microsoft, OpenAI or A…

A question, a typo, a bug?

This tour evolves with every model release. Your feedback is the raw material.

Contribute ← All tours