Shop by use

Local AI

Run language and image models on your own hardware: private, offline and with no monthly bill.

Buying tip: Memory decides which models fit. Speed comes second. Check the model size first, then pick the machine.

NVIDIA GeForce RTX 5090

NVIDIA · 32GB

NVIDIA GeForce RTX 5090

32 GB runs 30B-class models comfortably at 4-bit, but a 70B model will not fit on one card. For 70B, plan for two cards or a large unified-memory system.

$1,999
AMD Radeon AI PRO R9700 (32 GB)

AMD · 32GB

AMD Radeon AI PRO R9700 (32 GB)

Check that your tools support ROCm on your operating system before buying. PyTorch and llama.cpp do; some niche tools still assume CUDA.

$1,299
Intel Arc Pro B60

Intel · 24GB

Intel Arc Pro B60

Some partners put two B60 GPUs on one board (48 GB in total). Those dual cards need a motherboard slot that supports x8/x8 bifurcation, so check your board first.

$599.99
Beelink GTR9 Pro

Beelink · 128GB

Beelink GTR9 Pro

On Windows the GPU can be given up to 96 GB of the 128 GB; Linux lets you assign more. If you plan to run the largest models, plan for Linux.

$4,790
Minisforum MS-S1 Max

Minisforum · 128GB

Minisforum MS-S1 Max

The PCIe slot takes low-profile cards and runs at x4 lanes. It is ideal for a faster network card, not for adding a full-size graphics card.

$2,719 – $3,523
NVIDIA DGX Spark

NVIDIA · 128GB

NVIDIA DGX Spark

Its memory bandwidth (273 GB/s) is lower than a high-end graphics card's, so large dense models generate text more slowly than they would on an RTX 5090. Its strengths are capacity and the CUDA software stack.

$4,699 – $10,593
Apple Mac Studio (M3 Ultra)

Apple · 96GB

Apple Mac Studio (M3 Ultra)

macOS keeps part of the unified memory for the system, so plan for the GPU to use most of it, not all. Size the configuration with some headroom above the model.

$4,404 – $10,682
NVIDIA RTX PRO 6000 Blackwell (96 GB)

NVIDIA · 96GB

NVIDIA RTX PRO 6000 Blackwell (96 GB)

Pick the Max-Q Workstation Edition (300 W, blower) when you plan to stack several cards in one chassis; the 600 W Workstation Edition is the faster choice for a single card.

$14,594
Tenstorrent Blackhole p150a

Tenstorrent · 32GB

Tenstorrent Blackhole p150a

This is not a CUDA card. Check that the models you need are supported in Tenstorrent's open-source model library before you buy.

$1,399
Tenstorrent Wormhole n300

Tenstorrent · 24GB

Tenstorrent Wormhole n300

The 24 GB is split across two chips (12 GB each), so larger models have to be spread across both. Check that the model you want has a working multi-chip version.

$1,399
Apple Mac mini (M4 Pro)

Apple · 24GB

Apple Mac mini (M4 Pro)

Memory cannot be upgraded after purchase. If local AI is the goal, choose 48 GB or 64 GB now. With 24 GB, models above about 14B parameters are a squeeze.

$1,541 – $3,083
NVIDIA GeForce RTX 4090

NVIDIA · 24GB

NVIDIA GeForce RTX 4090

Push the 16-pin power cable fully into the card and avoid sharp bends close to the plug. A loose connector is the most common cause of trouble with this card.

$1,761
AMD Radeon PRO W7900 (48 GB)

AMD · 48GB

AMD Radeon PRO W7900 (48 GB)

Choose the W7900 when model or scene size matters most. The newer R9700 has less memory (32 GB) but adds FP8 support and is faster per watt for AI.

$3,999
AMD Radeon RX 7900 XTX

AMD · 24GB

AMD Radeon RX 7900 XTX

Partner cards are often longer and thicker than the reference design. Measure your case and check your power supply before ordering.

$990
Intel Arc Pro B50

Intel · 16GB

Intel Arc Pro B50

No power cable is needed, so it suits small-form-factor machines. 16 GB runs 7B to 14B models at 4-bit; anything larger needs a bigger card.

$329