Can I Run Ministral on RTX 3060 12 GB Desktop (12 GB VRAM, 32 GB RAM)?

Superseded model. Ministral has been superseded by Mistral Small 4.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 October 16, 2024

Yes

Yes — Ministral 8B at Q8_0 needs about 10.5 GB of the 12 GB usable on RTX 3060 12 GB Desktop (12 GB VRAM, 32 GB RAM) (~1.5 GB spare), at ~27.7 tok/s (estimated), with room for about 16,384 tokens of context.

Confidence: high · Recommended quantization: Q8_0 · Estimated speed: ~27.7 tok/s

RTX 3060 12 GB Desktop (12 GB VRAM, 32 GB RAM) — what it gives a model

Usable memory for models12 GB
Memory bandwidth360 GB/s
Form factorDesktop
Operating systemWindows or Linux
Memory upgradeableYes

Ministral on RTX 3060 12 GB Desktop (12 GB VRAM, 32 GB RAM): memory by quantization

QuantMemory neededFits 12 GB?Max contextEst. speedDownload
F1618 GB✗ No16 GB
Q8_010.5 GB✓ Yes16K~27.7 tok/s8.5 GB
Q6_K8.6 GB✓ Yes16K~34.7 tok/s6.6 GB
Q5_K_M7.7 GB✓ Yes32K~39.2 tok/s5.7 GB
Q4_K_M6.9 GB✓ Yes32K~44.6 tok/s4.8 GB
Q3_K_M5.4 GB✓ Yes32K~58.3 tok/s3.4 GB
Q2_K4.6 GB✓ Yes32K~70.2 tok/s2.6 GB

Which Ministral sizes fit

VariantNeeds at Q4_K_MFits?Est. speed
Ministral 8B6.9 GB✓ Fits~44.6 tok/s
Ministral 3B3.8 GB✓ Fits~84 tok/s

RTX 3060 12 GB desktop limitations

Recommended setup

Ollama or llama.cpp (CUDA/ROCm) — vLLM if you need concurrent serving

How these numbers are calculated

FAQ

Can I run Ministral on RTX 3060 12 GB Desktop (12 GB VRAM, 32 GB RAM)?

Yes — Ministral 8B at Q8_0 needs about 10.5 GB of the 12 GB usable on RTX 3060 12 GB Desktop (12 GB VRAM, 32 GB RAM) (~1.5 GB spare), at ~27.7 tok/s (estimated), with room for about 16,384 tokens of context.

Which quantization of Ministral should I use on RTX 3060 12 GB Desktop (12 GB VRAM, 32 GB RAM)?

Q8_0 — it needs about 10.5 GB of the 12 GB available, downloads as roughly 8.5 GB, and runs at an estimated 27.7 tokens/sec with up to 16K of context.

What limits Ministral on RTX 3060 12 GB Desktop (12 GB VRAM, 32 GB RAM)?

Nothing binding — the model fits with headroom and generates at a usable speed on this hardware.

Which runtime should I use?

Ollama or llama.cpp (CUDA/ROCm) — vLLM if you need concurrent serving

Other Computers

Other Models on RTX 3060 12 GB Desktop (12 GB VRAM, 32 GB RAM)

Ministral on GPUs

What This Model Is Good At

Model & Tools

← Can I Run It? | Ministral model page | Check your hardware