Qwen3 Max (Preview)

AlibabaQwen

发布日期

2025-09-05

参数规模

—

上下文长度

262K

支持模态

text

能力雷达图

general

coding

reasoning

science估算

agents

multimodal

Science 在缺少专门科学评测时使用推理能力代理估算。

排行榜排名

领域	#排名	分数	来源
代码能力榜	181	51.0	AA
通用能力榜	214	48.0	AA
数学推理	99	76.0	AA
科学能力	173	52.0	AA

基准测试分数 (LLM Stats)

3d

SUNRGBD

0.33 / 100自报

Hypersim

0.13 / 100自报

Agents

t2-bench

81.2%自报

AndroidWorld_SR

71.1%自报

BFCL-V4

67.3%自报

BrowseComp

61.0%自报

FullStackBench en

58.1%自报

WideSearch

57.1%自报

TIR-Bench

55.5%自报

FullStackBench zh

55.0%自报

OSWorld-Verified

54.5%自报

Terminal-Bench 2.0

40.5%自报

VITA-Bench

31.9%自报

DeepPlanning

22.8%自报

Biology

GPQA

84.2%自报

Chemistry

SuperGPQA

63.4%自报

Code

SWE-Bench Verified

69.2%自报

Communication

Multi-Challenge

60.0%自报

Embodied

EmbSpatialBench

0.83 / 100自报

Finance

MMLU-Pro

85.3%自报

MMLU-ProX

81.0%自报

General

MMLU-Redux

93.3%自报

IFEval

91.9%自报

C-Eval

90.2%自报

MAXIFE

86.6%自报

Global PIQA

86.6%自报

MMMLU

85.2%自报

MMStar

81.9%自报

MMMU

81.4%自报

Include

79.7%自报

MMMU-Pro

75.1%自报

LiveCodeBench v6

74.6%自报

IFBench

70.2%自报

LongBench v2

59.0%自报

SimpleVQA

0.58 / 100自报

NOVA-63

57.1%自报

Grounding

RefCOCO-avg

0.89 / 100自报

ScreenSpot Pro

68.6%自报

RefSpatialBench

0.64 / 100自报

Healthcare

VideoMMMU

80.4%自报

SlakeVQA

78.7%自报

PMC-VQA

62.0%自报

MedXpertQA

61.4%自报

Image To Text

OCRBench

91.0%自报

Language

LingoQA

79.2%自报

WMT24++

76.3%自报

Long Context

MLVU

85.6%自报

LVBench

71.4%自报

MMLongBench-Doc

0.59 / 100自报

AA-LCR

58.5%自报

Math

HMMT25

89.2%自报

HMMT 2025

89.0%自报

MathVista-Mini

86.2%自报

DynaMath

85.0%自报

MathVision

83.9%自报

CodeForces

0.82 / 3000自报

PolyMATH

64.4%自报

Humanity's Last Exam

47.4%自报

Multimodal

VLMsAreBlind

97.0%自报

92.7%自报

AI2D

92.6%自报

MMBench-V1.1

91.5%自报

OmniDocBench 1.5

89.3%自报

VideoMME w sub.

86.6%自报

VideoMME w/o sub.

82.5%自报

CC-OCR

80.7%自报

CharXiv-R

77.5%自报

MVBench

74.8%自报

MMVU

72.3%自报

BabyVision

38.4%自报

ZEROBench-Sub

0.34 / 100自报

Nuscene

14.6%自报

ZEROBench

0.08 / 100自报

Reasoning

CountBench

0.98 / 100自报

BrowseComp-zh

69.5%自报

Hallusion Bench

67.9%自报

ERQA

64.8%自报

Seal-0

41.4%自报

OJBench

36.0%自报

Spatial Reasoning

RealWorldQA

84.1%自报

Vision

ODinW

42.6%自报

AA 评测指数

Math Index

75.0

Intelligence Index

19.2

Mmlu Pro

0.8

Gpqa

0.8

Aime 25

0.8

Livecodebench

0.7

Ifbench

0.5

Lcr

0.4

Scicode

0.4

Tau2

0.3

Terminalbench Hard

0.2

Hle

0.1

LLM Stats 分类评分

Math

Physics

Structured Output

Image To Text

Instruction Following

Language

Legal

Embodied

Finance

General

Biology

Text-to-image

Video

Multimodal

Reasoning

Spatial Reasoning

Long Context

Frontend Development

Grounding

Healthcare

Chemistry

Vision

Code

Communication

Economics

Tool Calling

Agents

Spatial

定价

输入价格$1.2 / 1M tokens

输出价格$6 / 1M tokens

混合价格(3:1)$2.4 / 1M tokens

速度

Tokens/秒60.4

首Token延迟1.82s

首回答延迟1.82s

供应商价格排行

5 个供应商

最便宜: Alibaba (China)最贵: LLM Gateway

供应商输入输出

1Alibaba (China)最便宜

$0.861

$3.441

2Alibaba主要

$1.2

3Abacus

$1.2

4Vercel AI Gateway

$1.2

5LLM Gateway

$1.2

比较该模型在不同 API 供应商之间的定价。

外部链接

Artificial Analysis