跳转到主要内容

Qwen3.6 Plus

AlibabaQwenProprietary

描述

Qwen3.6 Plus is Alibaba's next-generation flagship model featuring a 1 million token native context window, up to 65,536 output tokens, and always-on chain-of-thought reasoning. It uses a next-generation hybrid architecture optimized for efficiency and scalability. It leads on Terminal-Bench 2.0 agentic coding (61.6), surpassing Claude 4.5 Opus, and achieves strong results on document understanding (OmniDocBench 91.2) and multimodal reasoning (MMMU 86.0). Compared to Qwen 3.5, it is significantly more decisive in reasoning, using fewer tokens on straightforward tasks with better agent stability.

发布日期
2026-04-02
参数规模
上下文长度
1.0M
支持模态
image, text, video

能力雷达图

45
general
43
coding
88
reasoning
59
science估算
60
agents
90
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
智能体与工具44
58.0
LS
代码能力榜31
78.0
AA
通用能力榜15
88.0
AA
多模态榜14
87.0
LS
推理能力28
82.0
LS
科学能力47
73.0
AA

基准测试分数 (LLM Stats)

Agents

WideSearch74.3%自报
MCP Atlas74.1%自报
TAU3-Bench70.7%自报
OSWorld-Verified62.5%自报
TIR-Bench61.6%自报
Terminal-Bench 2.061.6%自报
Claw-Eval58.7%自报
SWE-Bench Pro56.6%自报
MCP-Mark48.2%自报
SkillsBench45.7%自报
VITA-Bench44.3%自报
DeepPlanning41.5%自报
Toolathlon39.8%自报
NL2Repo37.9%自报

Biology

GPQA90.4%自报

Chemistry

SuperGPQA71.6%自报

Code

SWE-Bench Verified78.8%自报
SWE-bench Multilingual73.8%自报

Finance

MMLU-Pro88.5%自报
MMLU-ProX84.7%自报

General

MMLU-Redux94.5%自报
IFEval94.3%自报
C-Eval93.3%自报
Global PIQA89.8%自报
MMMLU89.5%自报
MAXIFE88.2%自报
LiveCodeBench v687.1%自报
MMMU86.0%自报
Include85.1%自报
MMStar83.3%自报
MMMU-Pro78.8%自报
IFBench74.2%自报
SimpleVQA0.67 / 100自报
LongBench v262.0%自报
NOVA-6357.9%自报

Grounding

RefCOCO-avg0.94 / 100自报
ScreenSpot Pro68.2%自报

Healthcare

VideoMMMU84.0%自报

Language

WMT24++84.3%自报

Long Context

MLVU86.7%自报
AA-LCR68.3%自报
MMLongBench-Doc0.62 / 100自报

Math

HMMT 202596.7%自报
AIME 202695.3%自报
HMMT2594.6%自报
We-Math89.0%自报
DynaMath88.0%自报
MathVision88.0%自报
HMMT Feb 2687.8%自报
IMO-AnswerBench83.8%自报
PolyMATH77.4%自报
Humanity's Last Exam28.8%自报

Multimodal

V*96.9%自报
AI2D94.4%自报
OmniDocBench 1.591.2%自报
Video-MME84.2%自报
CC-OCR83.4%自报
CharXiv-R81.5%自报

Reasoning

CountBench0.98 / 100自报
ERQA65.7%自报

Spatial Reasoning

RealWorldQA85.4%自报

Vision

ODinW51.8%自报

AA 评测指数

Intelligence Index
50.0
Coding Index
42.9
Tau2
1.0
Gpqa
0.9
Ifbench
0.8
Lcr
0.7
Terminalbench Hard
0.4
Scicode
0.4
Hle
0.3

LLM Stats 分类评分

Video
90
Biology
90
Language
90
Spatial Reasoning
80
Structured Output
80
Text-to-image
80
Vision
80
Chemistry
80
Finance
80
Frontend Development
80
General
80
Grounding
80
Healthcare
80
Instruction Following
80
Legal
80
Math
80
Multimodal
80
Physics
80
Reasoning
80
Code
70
Economics
70
Image To Text
70
Long Context
70
Search
70
Tool Calling
60
Agents
60
Coding
50

定价

输入价格$0.5 / 1M tokens
输出价格$3 / 1M tokens
混合价格(3:1)$1.125 / 1M tokens

速度

Tokens/秒52.7 tokens/s
首Token延迟1.69s
首回答延迟107.01s

可用提供商

(LS 内部计价单位)
提供商输入价格输出价格
Together500K3.0M

外部链接