FitMyLLM
Mistral AI/Dense

Mistral AIMinistral 3 3B Reasoning

Ministral 3 3B Reasoning — the thinking-tuned sibling of Ministral 3 3B. Same shape, reasoning post-train, slightly larger from the extra head.

chatreasoningmathmultilingual
4.25B
Parameters
256K
Context length
3
Benchmarks
6
Quantizations
121K
HF downloads
Architecture
Dense
Released
2025-12-02
Layers
26
KV Heads
8
Head Dim
128
Family
mistral

Quantization Options

Context length:
QuantBitsVRAM @ 16KQuality
Q4_K_M4.89
4.3 GB
3.1 + 1.2 KV
good
Q5_K_S5.57
4.7 GB
3.4 + 1.2 KV
good
Q5_K_M5.7
4.7 GB
3.5 + 1.2 KV
good
Q6_K6.56
5.2 GB
4.0 + 1.2 KV
excellent
Q8_08.5
6.2 GB
5.0 + 1.2 KV
lossless
FP1616
10.2 GB
9.0 + 1.2 KV
lossless

Select your GPU above to see speed estimates and compatibility for each quantization.

Deploying for a team or in production? Size GPUs, cost & scaling in Enterprise →
Ready to run this?Rent by the hour

Rent a GPU and run Ministral 3 3B Reasoning now

Spin up an A100 / H100 / 4090 in ~60s. Pay by the second. Cancel anytime.

Community Ratings

Loading ratings...

Benchmarks (3)

Run this model

Easiest way to get started·Beginners
Docs ↗
curl -fsSL https://ollama.com/install.sh | sh
$ollama run hf.co/unsloth/Ministral-3-3B-Reasoning-2512-GGUF:Q4_K_M

Not in Ollama's library: Ollama pulls the GGUF from Hugging Face (unsloth/Ministral-3-3B-Reasoning-2512-GGUF). If Q4_K_M is not in that repo, pick a file listed there.

Setup guide
>_

Auto-setup with fitmyllm CLI

Detects your GPU, recommends the best model, downloads it, and starts chatting — zero config. Benchmarks your speed and contributes anonymous data to improve predictions.

pip install fitmyllmthen run fitmyllmLearn more
Auto-detect GPULive tok/s in chatSpeed benchmarks9 inference engines

GPUs that can run this model

At Q4_K_M quantization. Sorted by minimum VRAM.

NVIDIA Tesla C1080
4 GB VRAM • 102 GB/s
NVIDIA
NVIDIA Tesla K10
4 GB VRAM • 160 GB/s
NVIDIA
NVIDIA Tesla M4
4 GB VRAM • 88 GB/s
NVIDIA
AMD Radeon Instinct MI8
4 GB VRAM • 512 GB/s
AMD
NVIDIA RTX A1000 Embedded
4 GB VRAM • 224 GB/s
NVIDIA
NVIDIA RTX A2000 Embedded
4 GB VRAM • 192 GB/s
NVIDIA

Find the best GPU for Ministral 3 3B Reasoning

Build Hardware for Ministral 3 3B Reasoning
Spec sheet

Ministral 3 3B Reasoning — 4.25B Dense.

Specifications
Parameters
4.25B
Architecture
Dense Transformer
Context Length
256K tokens
Capabilities
chat, reasoning, math, multilingual
Release Date
2025-12-02
Provider
Mistral AI
Family
mistral
VRAM requirements
QuantBPWVRAMQuality
Q4_K_M4.893.1 GB94%
Q5_K_S5.573.4 GB96%
Q5_K_M5.73.5 GB96%
Q6_K6.564.0 GB97%
Q8_08.55.0 GB100%
FP16169.0 GB100%
Benchmark scores
GPQA Diamond53.4
LiveCodeBench54.8
AIME72.1
Compatible GPUs
30 @ Q4_K_M