Book a 15-minute call
Learn
Blog and guidesToolsDev PlaygroundTemplatesFAQRegulation
Hardware Developers
For business
Local AI consultingEU AI Act: guide and templatesAI in your businessGrantsPublic sector
About
AboutPress
Contact Español
2026 catalogue

Hardware for every local AI job

From a €200 board to GPU servers, with the maths in plain sight. We do not sell hardware: we show you what runs your job, cheapest first.

Find the machine for your job

There is hardware for every job. Tell us the job and we show you what runs it, cheapest first.

Some links on this page are affiliate links: if you buy through them we may earn a commission, at no extra cost to you. It does not change what we recommend. How this works

Recommended computed, not hand-picked

Cheapest that runs a 7–8B model comfortably Beelink SER8 ~€550 Cheapest for 32B models Mac Mini M4 Pro (48GB) ~€1,840 Cheapest for 70B models Jetson AGX Thor (128GB) ~€3,499 Lowest power draw for a 7–8B model MacBook Pro M4 Pro (24GB) ~€2,199

Marks on the bar: what each model size needs at Q4 with 8k context. "Fits" is memory; "practical" also requires the machine's speed estimate to cover that size. Filters and picks use "practical". Indicative prices.

How we work with hardware

We do not sell hardware or get paid to recommend a brand. The machine is yours from day one; we size it, set it up and leave it running.

  1. A 15-minute call

    You tell us the job and the volume. We tell you whether you need a machine or an API is enough.

  2. Sized on your data

    We test the model on your tasks and tell you which machine runs it. You buy it, in your name, wherever you like.

  3. Setup and handover

    We set it up on your network, document it and train the people who will use it.

Book a 15-minute call

Compare two or three machines

Pick two or three to see them side by side.

See the full table
Machine Indicative price Memory Max draw Power/mo 24 h
RPi 5 + AI HAT+ 2 €200 16 GB 15 W €3
Jetson Orin Nano Super €230 8 GB 15 W €3
Beelink SER8 €550 32 GB 45 W €8
ASUS NUC 14 Pro €650 64 GB 28 W €5
Minisforum UM890 Pro €750 96 GB 45 W €8
Mac Mini M4 (24GB) €920 24 GB 30 W €5
RTX 4060 Ti 16GB Build €1,000 16 GB 300 W €54
RTX 3090 24GB (Used) Build €1,200 24 GB 500 W €90
Lenovo ThinkPad P16s Gen 3 (RTX A500) €1,650 32 GB 45 W €8
Mac Mini M4 Pro (48GB) €1,840 48 GB 40 W €7
Jetson AGX Orin 64GB €1,840 64 GB 60 W €11
Framework 16 (Radeon RX 7700S) €1,900 32 GB 40 W €7
Jetson T4000 (64GB) €1,999 64 GB 70 W €13
MacBook Pro M4 Pro (24GB) €2,199 24 GB 25 W €5
Mac Studio M4 Max (48GB) €2,300 48 GB 75 W €14
RTX 4090 24GB Build €2,500 24 GB 600 W €108
Jetson AGX Thor (128GB) €3,499 128 GB 130 W €23
AMD W7900 48GB Build €4,000 48 GB 400 W €72
Supermicro 1U Server (L40S 48GB) €12,000 48 GB 350 W €63
Dell PowerEdge R760xa (2x RTX A6000) €18,500 96 GB 600 W €108

Power at €0.25/kWh with the machine at maximum draw 24 hours a day for 30 days: a ceiling, not what normal use costs.

When does it beat the cloud?

Every machine page works out running cost and break-even with your query volume and verified API prices.

See a worked example VRAM explorer: which model fits ROI calculator
The EU AI Act now applies: a checklist you can complete

Tell us what you want to run

Tell us what you want to run and on what budget. We will tell you which hardware you need, which model fits, and what to expect from it, before you spend anything.

Let's talk for 15 minutes WhatsApp
First call free, 15 min Local-first: your data stays on your network Open tools and guides

69 free guides · 17 compliance templates