Mac miniAvailable
Mac mini M5 Pro
15-core CPU / 16-core GPU, 24GB
$1,699at Apple Store
24 GB · 307 GB/s · ~17 GB usable
Check priceBudget guide · Mini PCs
Best mini PC for local LLMs under $2,000
1 of the 6 priced mini PCs we track costs $2,000 or less right now. Top pick: the Mac mini M5 Pro at $1,699 with 24 GB (~17 GB usable for models), which runs up to GPT-OSS 20B at ~82 tok/s. Stretch a little and the GMKtec EVO-X2 at $2,200 gets you 64 GB. Prices & stock checked 2026-10-24. Prices change daily and some retailers block our checker, so always confirm on the store page.
Mini PCs under $2,000
Just over $2,000
Strix HaloAvailable
GMKtec EVO-X2
Ryzen AI Max+ 395, 64GB / 1TB
$2,200at GMKtec store
64 GB · 256 GB/s · ~48 GB usable
Check priceStrix HaloAvailable
GMKtec EVO-X2
Ryzen AI Max+ 395, 128GB / 1TB
$3,500at GMKtec store
128 GB · 256 GB/s · ~96 GB usable
Check price
Strix HaloAvailable
Minisforum MS-S1 Max
Ryzen AI Max+ 395, 128GB / 2TB
$3,919at Minisforum store
128 GB · 256 GB/s · ~96 GB usable
Check priceWhat each one runs (4-bit)
| Model | Mac mini M5 Pro$1,699 | GMKtec EVO-X2$2,200 | GMKtec EVO-X2$3,500 | Minisforum MS-S1 Max$3,919 |
|---|---|---|---|---|
| Qwen3.6 35B-A3B~24 GB at 4-bit | Won't fit | Fits~72 tok/s | Fits~72 tok/s | Fits~72 tok/s |
| GPT-OSS 20B~13 GB at 4-bit | Fits~82 tok/s | Fits~69 tok/s | Fits~69 tok/s | Fits~69 tok/s |
| Qwen3.8 27B~19 GB at 4-bit | Won't fit | Fits~12 tok/s | Fits~12 tok/s | Fits~12 tok/s |
| Llama 3.3 70B~46 GB at 4-bit | Won't fit | Tight fit~5 tok/s | Fits~5 tok/s | Fits~5 tok/s |
| GPT-OSS 120B~67 GB at 4-bit | Won't fit | Won't fit | Fits~55 tok/s | Fits~55 tok/s |
Estimated single-user decode at ~8K context. Mini PCs only: complete small-form-factor computers, no GPU card or host PC needed. For every rig under this price, including Mac Studio and GPU cards, see best local AI rig under $2,000. How we estimate
FAQ
What's the best mini PC for local LLMs under $2,000?
Of the mini PCs we track, the Mac mini M5 Pro at $1,699 (Apple Store, checked 2026-10-24) has the most usable memory for models in this budget (~17 GB). Just over budget, the GMKtec EVO-X2 at $2,200 has 64 GB.
How much memory does a mini PC need for local AI?
Memory decides which models fit. On the mini PCs we track at 4-bit: 24 GB (~17 GB usable) fits up to GPT-OSS 20B (biggest dense: Gemma 4 12B); 64 GB (~48 GB usable) fits up to Llama 3.3 70B; 128 GB (~96 GB usable) fits up to Qwen3.5 122B-A10B (biggest dense: Llama 3.3 70B). Unified-memory machines only let the GPU use about 70–75% of RAM by default.
Is a mini PC fast enough for LLMs?
Speed is set by memory bandwidth, 256–307 GB/s on the mini PCs we track. That's fine for mixture-of-experts models (we estimate 18–82 tok/s at 4-bit) but slow for big dense models (~5 tok/s on Llama 3.3 70B).
Mac mini or Strix Halo mini PC?
The Mac mini M5 Pro ($1,699) has more bandwidth (307 vs 256 GB/s) but 24 GB, so it's quicker on models that fit. Strix Halo boxes start at $2,200 with 64, 128 GB options, so they run much bigger models.