Mistral

Ministral 3 8B

Name: Ministral 3 8B
Rating: 81 (151 reviews)
Author: Mistral

Frontier

HuggingFace

187.1KDownloads172LikesOct 2025Veröffentlicht262K TokenKontextApache 2.0Lizenz72 StarkQualität

Ministral 3 8B (8B parameters) requires approximately 10.1 GB of VRAM with Q4_K_M quantization. For the best balance of quality and speed, we recommend hardware with at least 12 GB of VRAM.

Loslegen

— kopieren & einfügen, um lokal auszuführen

Copy-paste commands to run Ministral 3 8B on your machine.

Run

docker run --rm -it ghcr.io/ggerganov/llama.cpp:full \
  --hf-repo "mistralai/Ministral-3-8B-Instruct-2512" \
  --hf-file "Ministral-3-8B-Instruct-2512-Q4_K_M.gguf" \
  -c 4096 -ngl 99

Quick specs

Parameters8B

Architecturemultimodal

Context262K tokens

Modalitytext+vision

Min RAM3.1 GB

Rec. RAM4.9 GB (Q4_K_M)

LicenseApache 2.0

FamilyMinistral

✓ Vision✓ Chat

About this model

A balanced model in the Ministral 3 family, Ministral 3 8B is a powerful, efficient tiny language model with vision capabilities.

•8.4B Language Model
•0.4B Vision Encoder
•Vision: Enables the model to analyze images and provide insights based on visual content, in addition to text
•Multilingual: Supports dozens of languages, including English, French, Spanish, German, Italian, Portuguese, Dutch, Chinese, Japanese, Korean, Arabic
•System Prompt: Maintains strong adherence and support for system prompts

Verwandte Modelle

Schnellauswahl

Bestes BudgetA

Intel Arc B580 12GB~$249 — 48 tok/s

Beste GesamtwahlS

RTX 4070 Ti Super 16GB~$799 — 112 tok/s

Beste Hardware

Top-Empfehlungen für Ministral 3 8B

RTX 4070 Ti Super 16GBS

16 GB

RTX 4080 Super 16GBS

16 GB

RTX 4090 Laptop 16GBS

Dieses Modell ausführen

Ministral 3 8B on RTX 4070 Ti Super 16GB Ministral 3 8B on RTX 4080 Super 16GB Ministral 3 8B on RTX 4090 Laptop 16GB

Quantisierungsoptionen

VRAM-Schätzungen nach Quantisierungsstufe

No hardware detected — fit column shows raw VRAM estimates

Quant	Bits	VRAM	Quality	Fit
Q2_K	2	3.1 GB	Low	—
Q3_K_S	3	3.9 GB	Low	—
NVFP4	4	4.5 GB	Medium	—
Q4_K_M	4	4.9 GB	Medium	—
Q5_K_M	5	5.8 GB	High	—
Q6_K	6	6.6 GB	High	—
Q8_0	8	8.6 GB	Very High	—
F16	16	16.4 GB	Maximum	—

Quality benchmarks

Ministral 3 8B benchmark scores

Benchmark verified

Reasoning

MMLU-Pro—

GPQA Diamond39.9%

MATH-500—

ARC Challenge88.0%

Source: official · 2025-12-01

Hardware-Kompatibilität

Eignungsschätzungen für alle Hardware

Rechner öffnen

Computing compatibility...

Speicheraufschlüsselung

Reference: RTX 2060 6GB

Weights4.9 GB

KV Cache2.2 GB

Runtime2.4 GB

Headroom0.6 GB

Häufig gestellte Fragen

FAQ — Ministral 3 8B

Siehe auch

Quantisierungsleitfaden Bewertungsmethodik Rechner öffnen