Can I Run Mistral Family on 8 GB system RAM?

Superseded model. Mistral Family has been superseded by Mistral Small 4. This page is kept for reference; the newer family is a better starting point. View Mistral Small 4 →

Written by Jakub Rusinowski · Last updated January 28, 2025

Yes, but it is tight

Yes, but it is tight — Mistral NeMo 12B at Q2_K needs about 6.1 GB of the 6.4 GB usable on 8 GB system RAM, leaving only ~0.3 GB before the runtime starts swapping. Expect ~14.1 tok/s (estimated), with room for about 8,192 tokens of context.

Confidence: medium · Recommended quantization: Q2_K · Estimated speed: ~14.1 tok/s

8 GB system RAM — what it gives a model

Usable memory for models	6.4 GB
Memory bandwidth	90 GB/s

Mistral Family on 8 GB system RAM: memory by quantization

Quant	Memory needed	Fits 6.4 GB?	Max context	Est. speed	Download
F16	26.1 GB	✗ No	—	—	24 GB
Q8_0	14.9 GB	✗ No	—	—	12.8 GB
Q6_K	12 GB	✗ No	—	—	9.8 GB
Q5_K_M	10.6 GB	✗ No	—	—	8.5 GB
Q4_K_M	9.4 GB	✗ No	—	—	7.2 GB
Q3_K_M	7.3 GB	✗ No	—	—	5.1 GB
Q2_K	6.1 GB	✓ Yes	8K	~14.1 tok/s	3.9 GB

Which Mistral Family sizes fit

Variant	Needs at Q4_K_M	Fits?	Est. speed
Mistral Small 3 (24B)	17.2 GB	✗ Too large	—
Mistral NeMo 12B	9.4 GB	✗ Too large	—

What to watch out for

Only ~0.3 GB of headroom at Q2_K: a longer context or a second application can push this into swapping.
Q2_K is the only quantization that fits, and it is a heavily degraded one — expect noticeably worse output than the same model at Q4_K_M. A smaller model at Q4 is usually the better trade.
2 larger variants of Mistral Family do not fit and would need CPU offload or different hardware.
These figures assume CPU-only inference. Any discrete GPU, even an 8 GB one, will be several times faster for models that fit in its VRAM.

Recommended setup

llama.cpp (CPU build) or Ollama — both run without a GPU

How these numbers are calculated

Memory need = quantized weights + KV cache at 8,192 tokens (f16) + 0.8 GB runtime overhead.
6.4 GB of the 8 GB is treated as usable for model weights (80% — the rest is the OS and running applications).
DDR5-5600 dual channel at 89.6 GB/s peak. CPU decode is assumed to sustain 35% of that peak, because CPU inference is not purely bandwidth-bound — it also spends real time in compute and thread synchronisation. This figure is an assumption, not a fitted constant: no CPU measurement is in the calibration set.
CPU-only inference: no GPU is assumed. A GPU of any size will beat these figures substantially.
Throughput is a memory-bandwidth roofline estimate, not a measurement. It is labelled "estimated" everywhere it appears.
KV cache is computed from this model's published attention configuration.

FAQ