加载中...

Qwen3-32B（Qwen3-32B）详细信息 | 名称、简介、使用方法，开源情况，商用授权信息 | DataLearnerAI

Qwen3-32B

Name: Qwen3-32B
Availability: InStock
Author: 阿里巴巴

Release date: 2025-04-28更新于: 2025-07-30 11:34:592,392

Live demo GitHub Hugging Face

Parameters

320.0亿

Context length

128K

Chinese support

Supported

Reasoning ability

Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology

Model basics

Reasoning traces

Supported

Context length

128K tokens

Max output length

16384 tokens

Model type

推理大模型

Release date

2025-04-28

Model file size

64GB

MoE architecture

Total params / Active params

320.0B / N/A

Knowledge cutoff

No data

Inference modes

No mode data

Open source & experience

Code license

Apache 2.0

Weights license

Apache 2.0- 免费商用授权

GitHub repo

GitHub link unavailable

Hugging Face

https://huggingface.co/Qwen/Qwen3-32B

Live demo

https://chat.qwen.ai/

Official resources

Paper

Qwen3: Think Deeper, Act Faster

DataLearnerAI blog

重磅！阿里开源第三代千问大模型：Qwen3系列，最小仅6亿参数规模，最大2350亿参数规模大模型！

API details

API speed

4/5

💡Default unit: $/1M tokens. If vendors use other units, follow their published pricing.

Standard pricingStandard

Modality	Input	Output
Text	$0.7	$2.8

Benchmark Scores

综合评估

3 evaluations

Benchmark / mode

Score

Rank/total

GPQA Diamondthinking

68.40

85 / 144

LiveBenchnormal

62.78

40 / 52

GPQA Diamondnormal

53.30

115 / 144

数学推理

5 evaluations

Benchmark / mode

Score

Rank/total

MATH-500thinking

97.20

13 / 42

AIME 2024normal

81.40

25 / 62

AIME 2024thinking

81.40

25 / 62

AIME2025normal

72.90

65 / 102

AIME2025thinking

72.90

65 / 102

编程与软件工程

1 evaluations

Benchmark / mode

Score

Rank/total

LiveCodeBenchnormal

65.70

44 / 101

写作和创作

1 evaluations

Benchmark / mode

Score

Rank/total

Creative Writingthinking

17 / 22

Agent能力评测

1 evaluations

Benchmark / mode

Score

Rank/total

Aider-Polyglotthinking

21 / 22

Publisher

阿里巴巴

View publisher details

Model Overview

Qwen3-32B是阿里开源的第三代千问大模型，参数规模320亿，阿里还发布了一个同等参数规模的大模型，即Qwen3-30B-A3B，是MoE架构。

这两个模型总参数量差不多，但是每次推理激活的参数差了10倍，二者模型能力差不多水平。

DataLearner 官方微信

欢迎关注 DataLearner 官方微信，获得最新 AI 技术推送