
Model fit
Which machine runs which model
Memory decides which models fit. Pick a model size and quality, and see every machine and card we sell that can hold it.
30–34B at 4-bit
About 20 GB of GPU memory
The sweet spot for many people: noticeably smarter answers, and still a single 24–32 GB card at 4-bit.
Examples: Gemma 2 27B, Qwen2.5 32B
13 of 21 machines and cards can run it
ASRock Intel Arc Pro B60 Creator 24GB24 GBAED 2,199
SAPPHIRE PULSE AMD Radeon RX 7900 XTX 24GB24 GBAED 3,639
PowerColor AMD Radeon AI PRO R9700 32GB32 GBAED 4,769
Tenstorrent Wormhole n300s24 GBAED 5,139
Tenstorrent Blackhole p150a32 GBAED 5,139
NVIDIA GeForce RTX 409024 GBAED 6,469
NVIDIA GeForce RTX 509032 GBAED 7,339
Minisforum MS-S1 Max128 GB unifiedAED 9,989
AMD Radeon PRO W7900 (48 GB)48 GBAED 14,689
Apple Mac Studio (M3 Ultra)96 GB unifiedAED 16,169
NVIDIA DGX Spark128 GB unifiedAED 17,259
Beelink GTR9 Pro128 GB unifiedAED 17,589
NVIDIA RTX PRO 6000 Blackwell Workstation Edition (96 GB)96 GBAED 53,599
All numbers are approximate. We count about (parameters × bits ÷ 8) for the weights plus 20–30% for context and buffers, then treat 90% of a graphics card's memory and 75% of a unified-memory system's memory as usable. Long contexts, image models and batching need more. One card is counted on its own; several cards can be combined in one machine.
Not sure which one?
Tell the assistant what you want to run and your budget.