跳转到主要内容
AITier

Qwen3 30B A3B 2507 (Reasoning)

AlibabaQwen开源权重Apache 2.0 · 商用许可

描述

Qwen3-30B-A3B is a smaller Mixture-of-Experts (MoE) model from the Qwen3 series by Alibaba, with 30.5 billion total parameters and 3.3 billion activated parameters. Features hybrid thinking/non-thinking modes, support for 119 languages, and enhanced agent capabilities. It aims to outperform previous models like QwQ-32B while using significantly fewer activated parameters.

发布日期
2025-07-30
参数规模
30.5B
上下文长度
131K
支持模态
text

能力雷达图

31
general
35
coding
73
reasoning
45
science
70
agents
0
multimodal

排行榜排名

领域#排名分数来源
代码能力榜379
37.0
AA
通用能力榜352
39.0
AA
科学能力335
43.0
AA

基准测试分数 (LLM Stats)

(LLM Stats (zeroeval))

Chat

Multi-IF72.2%自报

General

Arena Hard91.0%自报
BFCL69.1%自报

Math

AIME 202480.4%自报
LiveBench74.3%自报
AIME 202570.9%自报

Reasoning

GPQANYU + Cohere + Anthropic (2023)65.8%自报
LiveCodeBench62.6%自报

AA 评测指数

(Artificial Analysis)
Math 500(OpenAI (2024), subset of Hendrycks et al. MATH (2021))
97.6
Aime(MAA (Mathematical Association of America))
90.7
Mmlu Pro(TIGER-Lab (Univ. of Waterloo, Toronto, CMU, 2024))
80.5
Gpqa(NYU + Cohere + Anthropic (2023))
70.7
Livecodebench(UC Berkeley + MIT + Cornell (2024))
70.7
Lcr(Artificial Analysis)
61.3
Aime 25(MAA (Mathematical Association of America))
56.3
Math Index(Artificial Analysis)
56.3
Ifbench(Google Research (2023))
50.7
Scicode(UIUC + Argonne National Lab (2024))
33.0
Tau2(Sierra + U Toronto + Vector Institute (2025))
28.1
Coding Index(Artificial Analysis)
12.1
Hle(Center for AI Safety + Scale AI (2025))
10.3
Intelligence Index(Artificial Analysis)
9.8
Tau Banking
5.4
Terminalbench Hard(Stanford × Laude Institute (2026))
5.3
Terminalbench V2 1
1.5

LLM Stats 分类评分

(LLM Stats (zeroeval))
Creativity
90
Writing
90
Chat
80
Math
80
Instruction Following
70
Language
70
Physics
70
Reasoning
70
Structured Output
70
General
70
Biology
70
Chemistry
70
Communication
70
Tool Calling
70
Code
60

定价

输入价格$0.2 / 1M tokens
输出价格$2.4 / 1M tokens
混合价格(3:1)$0.75 / 1M tokens

速度

Tokens/秒0.0
首Token延迟0.00s
首回答延迟0.00s

供应商价格排行

供应商价格排行

4 个供应商

最便宜: Novita最贵: Alibaba
供应商输入输出
1Novita最便宜
$0
$0
2DeepInfra
$0
$0
3Fireworks
$0
$0
4Alibaba主要
$0.2
$2.4

比较该模型在不同 API 供应商之间的定价。

外部链接

LLM StatsArtificial Analysis