Hardware · Memory

How much memory does a model need?

Pick an LLM and a context length. We add up the weights at every quantization, the KV cache that grows with context, and runtime overhead, then show the smallest systems and devices that hold it.

Model

Context length

Mistral Large 4 (1T-A52B) at 8K context

Quantization	Weights	KV cache + overhead	Total	Smallest hardware that holds it
`Q2_K` Low	421 GB	1.2 GB	422 GB	AMD EPYC 7002 / 7003 series NAS (512 GB) Intel Xeon D-2700 series NAS (512 GB) Intel Xeon Silver (4100 to 4300 series) NAS (512 GB) 4 in all
`Q3_K_M` Low	516 GB	1.2 GB	517 GB	Nothing in the catalog holds this on one machine.
`Q4_K_M` Good	642 GB	1.2 GB	644 GB	Nothing in the catalog holds this on one machine.
`Q5_K_M` High	748 GB	1.2 GB	749 GB	Nothing in the catalog holds this on one machine.
`Q6_K` High	864 GB	1.2 GB	865 GB	Nothing in the catalog holds this on one machine.
`Q8_0` Very high	1,116 GB	1.2 GB	1,118 GB	Nothing in the catalog holds this on one machine.
`F16` Full	2,106 GB	1.2 GB	2,108 GB	Nothing in the catalog holds this on one machine.

Hardware is listed when its usable memory is at least 8% above the total. Macs count roughly two-thirds to three-quarters of unified memory as usable. Each link opens the full “can it run” breakdown with speed.

The rule of thumb

1 · WeightsParameters × bytes eachA 70B model at 4 bits per weight is about 70 × 0.6 ≈ 42 GB. Lower quantization shrinks it and costs some quality.

2 · KV cacheGrows with contextEvery token you keep in context stores keys and values for each layer. Doubling context doubles this part, so long documents cost real memory.

3 · OverheadLeave headroomThe runtime, the operating system and other apps need memory too. Aim for a total at least 10% under the memory your GPU or device can give a model.

Estimates come from published hardware specs and our open formulas, not from running the model on your machine. Real speed varies with runtime, drivers and settings. Numbers marked “Measured” come from a linked benchmark.