Mistral

Magistral Small 2507

Name: Magistral Small 2507
Rating: 90 (95 reviews)
Author: Mistral

Legacy

HuggingFace

Ollama

1.7KDownloads104LikesJul 2025Veröffentlicht131K TokenKontextApache 2.0Lizenz96 HerausragendQualität

Magistral Small 2507 (24B parameters) requires approximately 18.9 GB of VRAM with Q4_K_M quantization. For the best balance of quality and speed, we recommend hardware with at least 22 GB of VRAM.

Loslegen

— kopieren & einfügen, um lokal auszuführen

Copy-paste commands to run Magistral Small 2507 on your machine.

Run

ollama run magistral

Quick specs

Parameters24B

Architecturedense

Context131K tokens

Modalitytext

Min RAM9.4 GB

Rec. RAM14.6 GB (Q4_K_M)

LicenseApache 2.0

FamilyMagistral

✓ Chat✓ Reasoning

About this model

Building upon Mistral Small 3.1 (2503), with added reasoning capabilities, undergoing SFT from Magistral Medium traces and RL on top, it's a small, efficient reasoning model with 24B parameters.

•Reasoning:: Capable of long chains of reasoning traces before providing an answer
•Multilingual:: Supports dozens of languages, including English, French, German, Greek, Hindi, Indonesian, Italian, Japanese, Korean, Malay,...
•Apache 2.0 License:: Open license allowing usage and modification for both commercial and non-commercial purposes
•Context Window:: A 128k context window, but performance might degrade past 40k. Hence we recommend setting the maximum model length to 40k

Verwandte Modelle

Schnellauswahl

Bestes BudgetS

Intel Arc Pro B60 24GB~$599 — 18 tok/s

Beste GesamtwahlS

RTX 5090 32GB~$1,999 — 88 tok/s

Beste Hardware

Top-Empfehlungen für Magistral Small 2507

RTX 5090 32GBS

32 GB

RTX PRO 4500 Blackwell 32GBS

32 GB

AMD Instinct MI100 32GBS

Dieses Modell ausführen

Magistral Small 2507 on RTX 5090 32GB Magistral Small 2507 on RTX PRO 4500 Blackwell 32GB Magistral Small 2507 on AMD Instinct MI100 32GB

Quantisierungsoptionen

VRAM-Schätzungen nach Quantisierungsstufe

No hardware detected — fit column shows raw VRAM estimates

Quant	Bits	VRAM	Quality	Fit
Q2_K	2	9.4 GB	Low	—
Q3_K_S	3	11.8 GB	Low	—
NVFP4	4	13.4 GB	Medium	—
Q4_K_M	4	14.6 GB	Medium	—
Q5_K_M	5	17.3 GB	High	—
Q6_K	6	19.7 GB	High	—
Q8_0	8	25.7 GB	Very High	—
F16	16	49.2 GB	Maximum	—

Quality benchmarks

Magistral Small 2507 benchmark scores

Benchmark verified

Reasoning

MMLU-Pro—

GPQA Diamond68.2%

MATH-50095.9%

ARC Challenge—

General

Chatbot Arena—

IFEval87.4%

Source: official · 2025-07-01

Hardware-Kompatibilität

Eignungsschätzungen für alle Hardware

Rechner öffnen

Computing compatibility...

Speicheraufschlüsselung

Reference: RTX 2060 6GB

Weights14.6 GB

KV Cache2.4 GB

Runtime1.2 GB

Headroom0.6 GB

Häufig gestellte Fragen

FAQ — Magistral Small 2507

Siehe auch

Quantisierungsleitfaden Bewertungsmethodik Rechner öffnen