Buy a box, or rent one?
14 reference machines people actually buy to run open models, against the cheapest VPS in the index with the same usable memory. Local hardware wins on speed (GPU-class bandwidth); a VPS wins on upfront cost and uptime. September 2026 — prices last checked 3 Sept 2026, refreshed weekly.
Reference machines for running open models locally. Memory, bandwidth and power come from vendor spec pages. Prices are refreshed weekly by scripts/update-hardware.ts: Apple prices are read from the Apple Store page (list price before tax; BTO memory upgrades add a curated delta, marked 'derived'); vendors that block automated fetches are checked through a web-search lookup and marked 'approximate' — those follow street prices, which in 2026 sit far above MSRP because of memory shortages. Prices are editable on the page — treat them as starting points.
| Machine | Price | Memory · usable | Bandwidth | Up to 4B | 7–9B | 12–15B | 24–32B | 70B+ | Mixture-of-experts | = VPS months |
|---|---|---|---|---|---|---|---|---|---|---|
| Raspberry Pi 5 (16 GB) Raspberry Pi · sbc · ~8 W | $305vendor list price · 2 Sept 2026 | 16 GB · 14.5 GB | 17 GB/s | – | – | 19 moContabo 24 GB · $16.27/mo | ||||
| GeForce RTX 5060 Ti 16 GB (card only) NVIDIA · card only · ~180 W | $805street, approx. · 2 Sept 2026 | 16 GB · 15.5 GB | 448 GB/s | – | – | 50 moContabo 24 GB · $16.27/mo | ||||
| Mac mini (M6, 16 GB) Apple · desktop · ~35 W | $899vendor list price · 3 Sept 2026 | 16 GB · 12 GB | 153 GB/s | – | – | – | 103 moContabo 12 GB · $8.72/mo | |||
| GeForce RTX 5070 Ti 16 GB (card only) NVIDIA · card only · ~300 W | $1,100street, approx. · 2 Sept 2026 | 16 GB · 15.5 GB | 896 GB/s | – | – | 68 moContabo 24 GB · $16.27/mo | ||||
| Mac mini (M6, 32 GB) Apple · desktop · ~35 W | $1,299vendor base + upgrade · 3 Sept 2026 | 32 GB · 24 GB | 153 GB/s | – | 80 moContabo 24 GB · $16.27/mo | |||||
| GeForce RTX 5080 16 GB (card only) NVIDIA · card only · ~360 W | $1,500street, approx. · 2 Sept 2026 | 16 GB · 15.5 GB | 960 GB/s | – | – | 92 moContabo 24 GB · $16.27/mo | ||||
| Mac mini (M5 Pro, 24 GB) Apple · desktop · ~50 W | $1,699vendor list price · 3 Sept 2026 | 24 GB · 18 GB | 307 GB/s | – | 104 moContabo 24 GB · $16.27/mo | |||||
| Framework Desktop (Ryzen AI Max+ 395, 64 GB) Framework · desktop · ~120 W | $1,959vendor list price · 2 Sept 2026 | 64 GB · 48 GB | 256 GB/s | 67 moContabo 48 GB · $29.05/mo | ||||||
| Mac mini (M5 Pro, 64 GB) Apple · desktop · ~55 W | $2,299vendor base + upgrade · 3 Sept 2026 | 64 GB · 48 GB | 307 GB/s | 79 moContabo 48 GB · $29.05/mo | ||||||
| Mac Studio (M5 Max, 36 GB) Apple · desktop · ~70 W | $2,499vendor list price · 3 Sept 2026 | 36 GB · 27 GB | 460 GB/s | – | 86 moContabo 48 GB · $29.05/mo | |||||
| Framework Desktop (Ryzen AI Max+ 395, 128 GB) Framework · desktop · ~120 W | $3,449vendor list price · 2 Sept 2026 | 128 GB · 96 GB | 256 GB/s | 30 moContabo 96 GB · $115.06/mo | ||||||
| NVIDIA DGX Spark (128 GB) NVIDIA · desktop · ~240 W | $4,699street, approx. · 2 Sept 2026 | 128 GB · 96 GB | 273 GB/s | 41 moContabo 96 GB · $115.06/mo | ||||||
| GeForce RTX 5090 32 GB (card only) NVIDIA · card only · ~575 W | $4,700street, approx. · 2 Sept 2026 | 32 GB · 31.5 GB | 1792 GB/s | – | 10+ yrsContabo 48 GB · $29.05/mo | |||||
| Mac Studio (M5 Ultra, 96 GB) Apple · desktop · ~110 W | $5,499vendor list price · 3 Sept 2026 | 96 GB · 72 GB | 1200 GB/s | 48 moContabo 96 GB · $115.06/mo |
Usable memory
Unified-memory machines (Apple, Strix Halo, DGX Spark) give the GPU about 75% of RAM by default; discrete cards lose ~0.5 GB to the driver; CPU-only boards keep 1.5 GB for the OS. Weights + KV cache must fit in what is left.
Why bandwidth decides speed
Generating a token reads every active weight once. An RTX 5090 at 1.8 TB/s streams an 8B Q4 model ~100× faster than a 4-vCPU VPS; a Mac mini sits in between. Per-model bands are on each model page.
2026 caveat
Memory-driven price rises have pushed GPUs 40–135% above MSRP and doubled the Raspberry Pi 5. Every model page has an editable calculator — plug in the price you can actually get.