Strix HaloAvailable
GMKtec EVO-X2
Ryzen AI Max+ 395, 64GB / 1TB
$2,200at GMKtec store
Cheapest that fits. Lowest price we've seen on any tracked rig that holds Qwen3.8 27B at 4-bit (19 of 48 GB usable).
Fits~12 tok/sQ4_K_M
Check priceAlibaba Qwen · 27B total / 27B active · dense
Best hardware to run Qwen3.8 27B locally
Qwen3.8 27B needs about 19 GB at Q4_K_M. 16 of the 17 priced rigs we track can hold it. Here's the cheapest, the best value and the fastest. Prices & stock checked 2026-10-23. Prices change daily and some retailers block our checker, so always confirm on the store page.
Our picks
Mac StudioAvailable
Mac Studio M5 Max
18-core CPU / 40-core GPU, 48GB
$3,099at Apple Store
Best value. Lowest price per token/sec among complete systems where it fits comfortably: about $96 per tok/s at 4-bit.
Fits~32 tok/sQ4_K_M
Suggested quant: Q8_0 (~19 tok/s)
Check price
RTX 5090In stock
NVIDIA GeForce RTX 5090 Founders Edition
32GB card
$8,500at Newegg
Fastest. Most memory bandwidth of the rigs that fit it (1,792 GB/s), so the quickest decode: ~83 tok/s at 4-bit.
Fits~83 tok/sQ4_K_M
GPU price is card only: you still need a host PC.
Check priceWhich quant to use
Use Q4_K_M on most rigs: it needs ~19 GB. Q8_0 (~32 GB) is a bit more accurate and fits comfortably on 3 of these rigs, e.g. the Mac Studio M5 Max.
Every rig that fits it
| Machine | Price | 4-bit fit & speed | Suggested quant | $ per tok/s |
|---|---|---|---|---|
| GMKtec EVO-X2Strix Halo | $2,200 | Fits~12 tok/s | Q4_K_M | $176 |
| Mac Studio M5 MaxMac Studio | $2,499 | Fits~25 tok/s | Q4_K_M | $102 |
| Mac Studio M5 MaxMac Studio | $3,099 | Fits~32 tok/s | Q8_0 | $96 |
| Framework Desktop DIYStrix Halo | $3,449 | Fits~12 tok/s | Q4_K_M | $276 |
| GMKtec EVO-X2Strix Halo | $3,500 | Fits~12 tok/s | Q4_K_M | $280 |
| Minisforum MS-S1 MaxStrix Halo | $3,919 | Fits~12 tok/s | Q4_K_M | $314 |
| Beelink GTR9 ProStrix Halo | $4,349 | Fits~12 tok/s | Q4_K_M | $348 |
| Gigabyte GeForce RTX 4090 Windforce V2RTX 4090 (card only, needs a host PC) | $4,899 | Fits~49 tok/s | Q4_K_M | $101 |
| Mac Studio M5 UltraMac Studio | $5,499 | Fits~55 tok/s | Q8_0 | $99 |
| Lenovo ThinkStation PGXDGX Spark / GB10 | $5,859 | Fits~13 tok/s | Q4_K_M | $456 |
| ASUS Ascent GX10DGX Spark / GB10 | $5,999 | Fits~13 tok/s | Q4_K_M | $467 |
| NVIDIA DGX SparkDGX Spark / GB10 | $6,950 | Fits~13 tok/s | Q4_K_M | $541 |
| HP ZGX Nano G1nDGX Spark / GB10 | $7,699 | Fits~13 tok/s | Q4_K_M | $600 |
| Gigabyte AI TOP ATOMDGX Spark / GB10 | $8,499 | Fits~13 tok/s | Q4_K_M | $662 |
| NVIDIA GeForce RTX 5090 Founders EditionRTX 5090 (card only, needs a host PC) | $8,500 | Fits~83 tok/s | Q4_K_M | $102 |
| PNY NVIDIA RTX PRO 6000 BlackwellRTX PRO 6000 (card only, needs a host PC) | $17,899 | Fits~83 tok/s | Q8_0 | $215 |
Sorted by lowest price. Speeds are estimated single-user decode at ~8K context. Full fit table, including rigs that don't fit: Qwen3.8 27B memory requirements. How we estimate
FAQ
How much memory does Qwen3.8 27B need?
About 19 GB at Q4_K_M and 32 GB at Q8_0, including runtime overhead and ~8K tokens of context for one user. Longer context needs more.
What's the cheapest hardware to run Qwen3.8 27B locally?
Of the rigs we track, the GMKtec EVO-X2 at $2,200, estimated ~12 tok/s at 4-bit. Prices checked 2026-10-23.
Which quant should I use for Qwen3.8 27B?
Q4_K_M is the default: about 19 GB. Q8_0 needs about 32 GB; use it only if you have the room and still get usable speed.
How fast does Qwen3.8 27B run?
It has 27B active parameters per token, so speed is set by memory bandwidth: from ~12 tok/s on the slowest rig that fits to ~83 tok/s on the NVIDIA GeForce RTX 5090 Founders Edition. Estimates, not benchmarks.