Magistral Medium 1

Mistral開源權重Apache 2.0 · 商用許可

描述

Trained solely with reinforcement learning on top of Mistral Medium 3, Magistral Medium is a reasoning model that achieves strong performance on complex math and code tasks without relying on distillation from existing reasoning models. The training uses an RLVR framework with modifications to GRPO, enabling improved reasoning ability and multilingual consistency.

發布日期

2025-06-10

參數規模

24.0B

上下文長度

128K

支援模態

text

能力雷達圖

general

coding

reasoning

science

agents

multimodal

排行榜排名

領域	#排名	分數	來源
程式碼能力榜	467	24.0	AA
通用能力榜	474	29.0	AA
科學能力	361	41.0	AA

基準測試分數 (LLM Stats)

(LLM Stats (zeroeval))

General

Aider-Polyglot

47.1%自報

Math

AIME 2024

73.6%自報

AIME 2025

64.9%自報

Reasoning

GPQANYU + Cohere + Anthropic (2023)

70.8%自報

LiveCodeBench

50.3%自報

Humanity's Last Exam

9.0%自報

AA 評測指數

(Artificial Analysis)

Math 500(OpenAI (2024), subset of Hendrycks et al. MATH (2021))

91.7

Mmlu Pro(TIGER-Lab (Univ. of Waterloo, Toronto, CMU, 2024))

75.3

Aime(MAA (Mathematical Association of America))

70.0

Gpqa(NYU + Cohere + Anthropic (2023))

67.9

Livecodebench(UC Berkeley + MIT + Cornell (2024))

52.7

Aime 25(MAA (Mathematical Association of America))

40.3

Math Index(Artificial Analysis)

40.3

Ifbench(Google Research (2023))

25.1

Tau2(Sierra + U Toronto + Vector Institute (2025))

23.1

Hle(Center for AI Safety + Scale AI (2025))

9.8

Intelligence Index(Artificial Analysis)

9.1

Terminalbench Hard(Stanford × Laude Institute (2026))

9.1

Lcr(Artificial Analysis)

0.0

LLM Stats 分類評分

(LLM Stats (zeroeval))

Physics

Biology

Chemistry

Math

Reasoning

General

Code

Vision

定價

輸入價格免費

輸出價格免費

混合價格(3:1)免費

速度

Tokens/秒0.0

首Token延遲0.00s

首回答延遲0.00s

供應商價格排行

1 個供應商

供應商輸入輸出

1Pioneer

比較該模型在不同 API 供應商之間的定價。

外部連結

LLM Stats Artificial Analysis