Software
Free
Open source, MIT license
Product guideSelf-hosted
Ollama
Run LLMs locally. Free, open-source. No API bills. Cost is hardware and power.
02 · Use cases
What teams actually use it for.
- Drafting and Q&A
- Code assistance
- Document analysis
- Private/internal use cases
- Offline or air-gapped
- Cost-predictable high volume
03 · Pricing
What you'd pay.
Free software. No per-token costs. You pay for: GPU/hardware, electricity, and maintenance.
Hardware
One-off
GPU required; 8GB+ VRAM for smaller models, 24GB+ for 14B, 40GB+ for 70B
Power
Ongoing
See our Self-hosted GPU Comparisons for £/1M token estimates
04 · Business fit
What to know before you commit.
It fits when
- 01 No data leaves your premises
- 02 Predictable cost (hardware + power)
- 03 No per-token bills
- 04 Wide model support (Llama, Mistral, Qwen, etc.)
- 05 Simple local setup
It doesn't fit when
- 01 Upfront hardware cost
- 02 You manage updates and security
- 03 Model quality depends on hardware
- 04 No managed SLA
When it makes sense
Good for data-sensitive workloads, high volume where cloud costs would exceed hardware, or air-gapped environments. Compare with our GPU cost guide.
05 · Data handling
Where your data goes.
Summary
All data stays on your hardware. No data sent to third parties. Full control.
GDPR · Compliance
Data never leaves your infrastructure.
Sovereignty
Complete data sovereignty.
Reference
Next step
Want a recommendation for your workload?
Every team's fit is different. We model cost and ROI across cloud, self-hosted, and hybrid before recommending anything. Including this product.
Next step
Book a demo. Get the numbers back.
A real workload, two sentences. We price it across cloud, self-hosted, and hybrid, then tell you what we'd run — or tell you to wait.
No deckNo sales pressureWe'll tell you to wait if you should
Or email a workload to hello@althorizon.co.uk — one-page model back in 48h.
