YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
Zero Point Intelligence Ltd β Getting Started
New to local AI? Start here.
Running AI on your own computer means no subscriptions, no data leaving your machine, no censorship, no limits. Here's how to get started in 5 minutes.
Step 1: What You Need
- A computer (Windows, Mac, or Linux β any modern one works)
- Ollama β free software that runs AI models β Download here
- A MARTHA model β pick one from our repos below
That's it. No Python. No coding. No cloud accounts.
Step 2: Install Ollama
Go to https://ollama.com and click Download. Install it like any normal app.
Step 3: Pick Your Model
| Your Hardware | Best Model | VRAM Needed | Download Size |
|---|---|---|---|
| Phone / Raspberry Pi | MARTHA-0.8B | 1 GB | ~500 MB |
| Laptop / Old PC | MARTHA-2B | 2 GB | ~1.5 GB |
| Gaming PC | MARTHA-4B | 4 GB | ~2.7 GB |
| Beefy GPU (RTX 3080+) | MARTHA-9B | 8 GB | ~5 GB |
Not sure about your VRAM? Start with MARTHA-4B. It runs on almost anything with a GPU.
Step 4: Download and Run
Go to the model page and follow the Quick Start instructions. Each model has copy-paste commands for:
- Windows (PowerShell)
- Mac/Linux (Terminal)
Step 5: Want a Pretty Chat Interface?
Install Docker Desktop, then run:
docker run -d -p 3000:8080 -v open-webui:/app/backend/data -e OLLAMA_BASE_URL=http://host.docker.internal:11434 --name open-webui ghcr.io/open-webui/open-webui:main
Open http://localhost:3000 in your browser. It gives you a web chat interface but runs entirely on your machine.
What Makes MARTHA Different?
Every MARTHA model ships with two files:
- The brain (
.gguf) β handles text, code, reasoning - The eyes (
mmproj-f16.gguf) β handles images, screenshots, photos
Most AI models on HuggingFace can only read text. MARTHA sees images too.
Jargon Buster
| You'll See This | It Means |
|---|---|
| GGUF | File format for AI models β like .mp4 for videos |
| Q4, Q5, Q8 | Compression levels β Q4 = small, Q8 = big but smarter |
| mmproj | The vision file β gives the model eyes |
| Ollama | Free app that runs AI locally |
| VRAM | GPU memory β more = bigger models |
| Safetensors | Raw model weights (for developers, not needed for basic use) |
| LoRA | A training technique β how we customise the model's personality |
| Ghost pass | Our method for creating unique model weights |
About Us
Zero Point Intelligence Ltd β Independent AI research lab, Dundee, Scotland π΄σ §σ ’σ ³σ £σ ΄σ Ώ
We build open-source AI models that run on your hardware. No cloud. No gatekeepers. No bullshit.
- Website: https://zeropointai.uk
- Email: zero.point.intelligence.ltd@zeropointai.uk
- HuggingFace: https://huggingface.co/ZERO-POINT-INTELLIGENCE-LTD
Intelligence From The Void