← All machines

Multi-GPU · DIY build

2x Intel Arc Pro B60

The cheapest new 48 GB build with a warranty: two 24 GB Battlemage workstation cards at 200 W each. Bandwidth (456 GB/s per card) is half a 3090's, so expect slower token generation.

Memory for models

46.5 GB

VRAM across all cards

Memory bandwidth

456 GB/s

per card; 2 cards take turns on each token

Parts

~$4,367

sum of parts

Largest at 4-bit

Llama 3.3 70B Instruct

43.6 GB at 8,192 tokens of context

gpt-oss-20b, 4-bit

~57–83 tok/sestimate

no measurement on this hardware yet

Parts

PartQtyEach
GPUIntel Arc Pro B60 24GB2~$599
CPUAMD Ryzen 7 9700X (8-core, 65 W)1~$359
MotherboardASUS ProArt X870E-CREATOR WIFI (2x PCIe 5.0 x16 slots, x16 or x8/x8)1~$550
RAMCorsair Vengeance 64GB (2x32GB) DDR5-6000 CL301~$1,275
PSUCorsair RM1000x (1000 W)1~$210
CaseCorsair 7000D AIRFLOW full tower (450 mm GPU clearance, 8 slots)1~$290
StorageCorsair MP600 CORE XT 2TB PCIe 4.0 NVMe1~$485
Total$4,367

New parts are priced at the maker’s own store or list price. Used cards have no price here yet, and a total with a missing GPU would understate the build, so none is shown.

What it runs at 4-bit

Every model in the catalog at its 4-bit file with 8,192 tokens of context. 21 of 41 fit. Speeds are the median of measured runs on the same GPUs where any exist; otherwise an estimate from memory bandwidth. Try other quants and context lengths.

ModelNeedsFitGeneration speed
Kimi K32.8T, 104B active · UD-Q4_K_XL1406.8 GBcontext memory estimatedDoes not fit
DeepSeek V4 Pro1.6T, 49B active · Q4_K_M890.5 GBcontext memory not countedDoes not fit
Kimi K2.61T, 32B active · UD-Q4_K_XL545.7 GBcontext memory estimatedDoes not fit
DeepSeek V4.1 Flash552B, 16B active · Q4 (DwarfStar)484.5 GBDoes not fit
GLM-5.3753.3B · UD-Q4_K_XL437.4 GBcontext memory estimatedDoes not fit
DeepSeek V3.2671B, 37B active · Q4_K_M379.6 GBcontext memory estimatedDoes not fit
MiniMax M3428B, 23B active · Q4_K_M244.8 GBcontext memory not countedDoes not fit
Qwen3.5 397B-A17B397B, 17B active · Q4_K_M229.1 GBcontext memory estimatedDoes not fit
GLM-5.3 Flash320B, 18B active · UD-Q4_K_XL187.6 GBcontext memory estimatedDoes not fit
Hy3295B, 21B active · Q4_K_M173.6 GBDoes not fit
MiniMax M2.7228.7B · Q4_K_M132.7 GBDoes not fit
Qwen3.8 Flash-Next125B, 6B active · Q4_K_M113.1 GBcontext memory estimatedDoes not fit
Nemotron 3 Super 120B-A12B120B, 12B active · Q4_K_M82.6 GBcontext memory estimatedDoes not fit
Qwen3.5 122B-A10B122B, 10B active · Q4_K_M73.0 GBcontext memory estimatedDoes not fit
Mistral Medium 3.5 128B128B · Q4_K_M74.0 GBDoes not fit
GLM-4.5-Air106B, 12B active · Q4_K_M70.9 GBDoes not fit
Mistral Small 4 119B (2603)119B, 6.5B active · Q4_K_M69.3 GBcontext memory estimatedDoes not fit
Llama 4 Scout 17B-16E Instruct109B, 17B active · Q4_K_M62.4 GBcontext memory not countedDoes not fit
gpt-oss-120b117B, 5.1B active · MXFP4 (native)60.2 GBcontext memory estimatedDoes not fit
Qwen3 Coder Next80B, 3B active · Q4_K_M46.9 GBcontext memory estimatedDoes not fit
Llama 3.3 70B Instruct70B · Q4_K_M43.6 GBFits~5.3–8.3 tok/sestimate
Nemotron 3.5 Lightning 30B-A3B30B, 3B active · Q4_K_M25.3 GBcontext memory estimatedFits~45–65 tok/sestimate
Qwen3.6 35B-A3B35B, 3B active · Q4_K_M22.4 GBcontext memory estimatedFits~59–87 tok/sestimate
Ornith 1.5 35B-A3B35B, 3B active · Q4_K_M21.9 GBcontext memory estimatedFits~61–89 tok/sestimate
Qwen3 Coder 30B-A3B Instruct30.5B, 3.3B active · Q4_K_M19.5 GBFits~57–83 tok/sestimate
Gemma 4 31B IT30.7B · Q4_K_M18.9 GBcontext memory estimatedFits~12–19 tok/sestimate
Granite 4.2 30B29.3B · Q4_K_M20.3 GBFits~12–19 tok/sestimate
Qwen3.8 27B27B · Q4_K_M18.2 GBcontext memory estimatedFits~13–20 tok/sestimate
Gemma 4 26B-A4B IT25.2B, 3.8B active · Q4_K_M17.4 GBcontext memory estimatedFits~44–64 tok/sestimate
Qwen3.6 27B27B · Q4_K_M17.7 GBcontext memory estimatedFits~13–21 tok/sestimate
Devstral Small 2 24B (2512)24B · Q4_K_M16.1 GBFits~16–24 tok/sestimate
gpt-oss-20b21B, 3.6B active · MXFP4 (native)12.5 GBcontext memory estimatedFits~57–83 tok/sestimate
Phi-414.7B · Q4_K_M11.3 GBFits~25–39 tok/sestimate
Ministral 3 14B Reasoning (2512)13.9B · Q4_K_M10.4 GBFits~27–43 tok/sestimate
Gemma 4 12B IT11.95B · Q4_K_M8.2 GBcontext memory estimatedFits~31–49 tok/sestimate
Qwen3.5 9B9B · Q4_K_M7.0 GBcontext memory estimatedFits~39–62 tok/sestimate
Granite 4.2 8B8.8B · Q4_K_M7.9 GBFits~40–63 tok/sestimate
Gemma 4 E4B IT8B · Q4_K_M6.5 GBcontext memory not countedFits~74–115 tok/sestimate
Qwen3 8B8.2B · Q4_K_M7.3 GBFits~44–70 tok/sestimate
Llama 3.1 8B Instruct8B · Q4_K_M7.1 GBFits~45–71 tok/sestimate
Qwen3.5 4B4B · Q4_K_M4.3 GBcontext memory estimatedFits~82–128 tok/sestimate

Measured speeds

Nobody has published a run on this GPU combination with a stated model, quant and backend that we could find. The speeds above are estimates.

Specifications

GPUs
2 × Intel Arc Pro B60
VRAM
48 GB
System RAM
64 GB
RAM bandwidth
2 channels x 6000 MT/s x 8 bytes = 96.0 GB/s (dual-channel AM5)
Power supply
1000 W
Estimated load
2x Arc Pro B60 200 W + Ryzen 7 9700X 65 W + 100 W = 565 W
Software
oneAPI/SYCL, Vulkan

Before you build it

  • The B60 is a native PCIe 5.0 x8 card, so the board's x8/x8 split costs it nothing.
  • Software is the trade-off: llama.cpp runs through its SYCL or Vulkan backend, and Intel's multi-GPU vLLM path ships as Linux containers.
  • The Maxsun dual-B60 card (48 GB on one board) needs a slot that bifurcates x16 into x8/x8, or the second GPU does not appear.
  • DDR5 prices are at shortage highs: Corsair lists a 64 GB (2x32) DDR5-6000 kit at $1,274.99, so RAM can cost more than a used GPU.

Sources

Compared with

Similar machines