Mistral AI

Mistral Nemo 12B

Mistral AI·12B·Mistral·1000K ctx·2026-08

Check on my hardwareAdd to compare →Ollama →Hugging Face →

Model size→Hardware requirement→Actual fit

Mistral Nemo 12B needs ~8.3–14.8 GB of VRAM depending on quantization. The bar is what it needs; the marker is the card's VRAM. Speed figures below are planning estimates.

8 GB

RTX 4060 · 3060 Ti

8.3
8
Q4_K_M · Est. ~3 tok/sSlow · CPU offload

12 GB

RTX 4070 · 3060 12G

9.6
12
Q5_K_M · Est. ~33 tok/sComfortable

16 GB

RTX 4080 · M-series 16G

9.6
16
Q5_K_M · Est. ~51 tok/sComfortable

24 GB

RTX 4090 · 3090

14.8
24
Q8_0 · Est. ~50 tok/sComfortable

48 GB

A6000 · 2×24G

14.8
48
Q8_0 · Est. ~50 tok/sComfortable

02  — Quantizations

QuantFile sizeEst. VRAM
Q4_K_M7.1 GB8.3 GB
Q5_K_M8.4 GB9.6 GB
Q8_013.6 GB14.8 GB

03  — Runtimes & use

llama.cpp · Ollama · LM Studio

General · Writing · Privacy

Memory and speed figures are planning estimates, not measured benchmarks. Memory includes a fixed allowance for KV cache and runtime overhead. Long-context prompts push VRAM use above the numbers shown here.