Skip to content
localmodel.run

Text model · Mistral

Ministral 3 3B: RAM and VRAM requirements

Ministral 3 3B needs about 3.2 GB to run at Q4_K_M (the Q4_K_M GGUF file is ~2 GB to download; KV cache and overhead add the rest), or about 4.6 GB at Q8_0. The lightest hardware that runs it is Nvidia GeForce RTX 2060 (6GB).

Mistral family · 3B params · released Dec 2025 · 1.4M Ollama pulls.

Shopping for hardware? See what runs Ministral 3 3B →

License Apache-2.0 · Commercial OK ↓ 311.1K/mo ♥ 278 on HuggingFace
Q4_K_M GGUF
2 GB
Q8_0 GGUF
3.4 GB
Memory @ Q4 (4k)
~3.2 GB
Context
256 k

Q4_K_M and Q8_0 are GGUF download (file) sizes. Memory @ Q4 includes the KV cache and overhead, so it runs larger than the file on disk.

Will it run on your device?

Ministral 3 3B runs on 43 of 43 tracked devices at Q4_K_M.

43 run well 0 tight fit 0 too small
Fit check Q4_K_M
Yes, it runs fast
needs 3.2 GB usable 10.5 GB
See the full breakdown
$ollama run ministral-3:3b

Memory breakdown

Weights (Q4_K_M)2 GB
+
KV cache (4k)0.4 GB
+
Overhead0.8 GB
=
Total3.2 GB

How context length changes it

4k context ~3.2 GB 32k context ~6.1 GB 128k context ~16.1 GB

Ministral 3 3B is small enough that the context window is the thing to size for: 128k needs ~16.1 GB versus ~3.2 GB at 4k, so pick the window you actually use.

Speed drops too: every token re-reads the KV cache, so at 128k context Ministral 3 3B generates at roughly ~13% of its short-context speed. How this is estimated.

Quantization sizes

GGUF quantson disk
Quantization Size on disk
Q2_K 1.3 GB est
Q3_K_M 1.5 GB est
Q4_K_M (default) 2 GB
Q5_K_M 2.1 GB est
Q6_K 2.5 GB est
Q8_0 3.4 GB
FP16 7.71 GB

Lower quant = smaller and faster, slightly lower quality. Q4_K_M is the common default.

Run it

Ollama
$ ollama run ministral-3:3b
llama.cpp
$ llama-cli -hf mistralai/Ministral-3-3B-Instruct-2512-GGUF:Q4_K_M
LM Studio
$ lms get mistralai/Ministral-3-3B-Instruct-2512-GGUF

Which devices can run Ministral 3 3B?

NVIDIA GPUs

Apple Silicon Macs

RAM-only laptops

iPhone & iPad

Android

AMD GPUs

Same job, different size

Pick by what fits your memory: step down to free up VRAM, or step up if you have headroom.

Next size down Gemma 2 2B 2.61B Next size up Qwen2.5 3B 3.09B

Similar models

Head-to-head

FAQ

How much VRAM or RAM does Ministral 3 3B need?

At Q4_K_M, Ministral 3 3B needs about 3.2 GB (weights ~2 GB + KV cache + overhead) at a 4k context. At Q8_0 budget ~4.6 GB.

What is the Q4_K_M GGUF file size of Ministral 3 3B?

The Q4_K_M GGUF file is about 2 GB to download, and the Q8_0 GGUF is about 3.4 GB. That is the weights file on disk; to run it you also need room for the KV cache and overhead, so budget ~3.2 GB of memory at Q4_K_M.

Can Ministral 3 3B run on a laptop?

Yes, Ministral 3 3B fits on a 16 GB machine at Q4_K_M and runs on Apple Silicon or a 12 GB+ GPU comfortably.

Can I use Ministral 3 3B commercially?

Yes. Ministral 3 3B is licensed Apache-2.0, which permits commercial use.

Understand the numbers

Short guides to the ideas behind Ministral 3 3B's memory and quant figures.

Mistral's edge family, smallest size: dense 3B with vision, 256K context, Apache-2.0. Sizes from Mistral's own GGUF repo. A phone and small-SBC class model.

Sources

Last validated 2026-09-07. Memory figures are estimates. See methodology.