Loading
Loading
Prices are our ballpark as of 2026-09-27, not a live store. Memory is what the planner uses.
One DGX Spark holds about a 200 billion parameter model after it is compressed. Two is the size most people finish. Four reaches the largest class and usually needs a fast network switch.
3 of 3 machines
Memory
128 GB coherent
Price
$4,700–6,500
What it can run
About a 200B-class quantized model. A Qwen ~35B is the calmer first success.
Best for
One serious local box. Not a cluster yet.
Memory
256 GB coherent
Price
$10–13k
What it can run
The sweet spot. DeepSeek V4 Flash and GLM 5.3 Flash across two nodes.
Best for
A local coding model you will finish setting up.
Memory
512 GB coherent
Price
$20–26k, plus a switch
What it can run
~700B-class. Full GLM 5.3 territory.
Best for
A lab that will own the network.