跳轉到主要內容

Ministral 3 3B

MistralMistral開源權重Apache 2.0 · 商用許可

描述

The smallest model in the Ministral 3 family, Ministral 3 3B is a powerful, efficient tiny language model with vision capabilities. This model is the base pre-trained version, not fine-tuned for instruction or reasoning tasks, making it ideal for custom post-training processes. The Ministral 3 family is designed for edge deployment, capable of running on a wide range of hardware. Ministral 3 3B can even be deployed locally, fitting in 16GB of VRAM in BF16, and less than 8GB of RAM/VRAM when quantized.

發布日期
2025-12-02
參數規模
3.0B
上下文長度
支援模態
image, text

能力雷達圖

20
general
13
coding
24
reasoning
22
science
21
agents
10
multimodal

排行榜排名

領域#排名分數來源
程式碼能力榜477
12.0
AA
通用能力榜464
24.0
AA
科學能力488
21.0
AA

基準測試分數 (LLM Stats)

(LLM Stats (zeroeval))

Biology

GPQANYU + Cohere + Anthropic (2023)53.4%自報

Code

LiveCodeBench54.8%自報

Communication

Wild Bench56.8%自報
MM-MT-Bench0.08 / 100自報

Creativity

Arena Hard30.5%自報

Finance

MMLU70.7%自報

General

MMLU-Redux73.5%自報
Multilingual MMLU65.2%自報
TriviaQA59.2%自報
AGIEval51.1%自報

Math

MATH83.0%自報
AIME 202477.5%自報
AIME 202572.1%自報
MATH (CoT)60.1%自報

AA 評測指數

(Artificial Analysis)
Math Index(Artificial Analysis)
22.0
Intelligence Index(Artificial Analysis)
7.1
Coding Index(Artificial Analysis)
4.8
Mmlu Pro(TIGER-Lab (Univ. of Waterloo, Toronto, CMU, 2024))
0.5
Gpqa(NYU + Cohere + Anthropic (2023))
0.4
Ifbench(Google Research (2023))
0.3
Tau2(Sierra + U Toronto + Vector Institute (2025))
0.2
Livecodebench(UC Berkeley + MIT + Cornell (2024))
0.2
Aime 25(MAA (Mathematical Association of America))
0.2
Lcr(Artificial Analysis)
0.2
Scicode(UIUC + Argonne National Lab (2024))
0.1
Hle(Center for AI Safety + Scale AI (2025))
0.1
Tau Banking
0.0
Terminalbench Hard(Stanford × Laude Institute (2026))
0.0
Terminalbench V2 1
0.0

LLM Stats 分類評分

(LLM Stats (zeroeval))
Language
70
Finance
70
Healthcare
70
Legal
60
Math
60
Reasoning
60
General
60

定價

輸入價格$0.1 / 1M tokens
輸出價格$0.1 / 1M tokens
混合價格(3:1)$0.1 / 1M tokens

速度

Tokens/秒202.4
首Token延遲0.47s
首回答延遲0.47s

供應商價格排行

供應商價格排行

4 個供應商

最便宜: Mistral最貴: Kilo Gateway
供應商輸入輸出
1Mistral主要
$0.1
$0.1
2NanoGPT
$0.1
$0.1
3OpenRouter
$0.1
$0.1
4Kilo Gateway
$0.1
$0.1

比較該模型在不同 API 供應商之間的定價。

外部連結