跳转到主要内容
AITier

Qwen3.5 27B (Reasoning)

AlibabaQwen开源权重Apache 2.0 · 商用许可

描述

Qwen3.5-27B is a multimodal dense foundation model with 27 billion parameters. It combines strong reasoning, coding, multilingual, long-context, and visual understanding performance in a production-friendly open-weight package with a native 262K context window.

发布日期
2026-02-24
参数规模
27.0B
上下文长度
262K
支持模态
audio, image, text, video

能力雷达图

23
general
60
coding
86
reasoning
67
science
60
agents
80
multimodal

排行榜排名

领域#排名分数来源
智能体能力模型榜57
48.0
LS
代码能力榜191
69.0
AA
通用能力榜124
64.0
AA
多模态榜74
53.0
LS
科学能力170
63.0
AA

基准测试分数 (LLM Stats)

(LLM Stats (zeroeval))

Agents

t2-bench79.0%自报
VITA-Bench41.9%自报
DeepPlanning22.6%自报

Chat

IFEvalGoogle Research (2023)95.0%自报
Multi-Challenge60.8%自报

Code

FullStackBench en60.1%自报
FullStackBench zh57.4%自报

General

C-Eval90.5%自报
MAXIFE88.0%自报
Include81.6%自报
AndroidWorld_SR64.2%自报
NOVA-6358.1%自报

Healthcare

MedXpertQA62.4%自报
PMC-VQA62.4%自报

Instruction Following

IFBench76.5%自报

Language

MMLU-Redux93.2%自报
MMLU-Pro86.1%自报
MMMLU85.9%自报
MMLU-ProX82.2%自报
WMT24++77.6%自报

Long Context

LongBench v260.6%自报

Math

HMMT 202592.0%自报
HMMT2589.8%自报
MathVista-Mini87.8%自报
DynaMath87.7%自报
MathVision86.0%自报
CodeForces0.81 / 3000自报
PolyMATH71.2%自报

Multimodal

VideoMME w/o sub.82.8%自报
VideoMMMU82.3%自报
MMMU82.3%自报
TIR-Bench59.8%自报
OSWorld-Verified56.2%自报

Reasoning

Global PIQA87.5%自报
GPQANYU + Cohere + Anthropic (2023)85.5%自报
LiveCodeBench v680.7%自报
CharXiv-R79.5%自报
SWE-Bench Verified72.4%自报
AA-LCR66.1%自报
SuperGPQA65.6%自报
BrowseComp-zh62.1%自报
BrowseCompOpenAI (2025)61.0%自报
Humanity's Last Exam48.5%自报
Seal-047.2%自报
Terminal-Bench 2.0Stanford × Laude Institute (2026)41.6%自报
OJBench40.1%自报

Search

WideSearch61.1%自报

Tool Calling

BFCL-V468.5%自报

Video

MLVU85.9%自报

Vision

CountBench0.98 / 100自报
VLMsAreBlind96.9%自报
V*93.7%自报
AI2D92.9%自报
MMBench-V1.192.6%自报
RefCOCO-avg0.91 / 100自报
OCRBench89.4%自报
OmniDocBench 1.588.9%自报
VideoMME w sub.87.0%自报
EmbSpatialBench0.84 / 100自报
RealWorldQA83.7%自报
LingoQA82.0%自报
CC-OCR81.0%自报
MMStar81.0%自报
SlakeVQA80.0%自报
MMMU-Pro75.0%自报
MVBench74.6%自报
LVBench73.6%自报
MMVU73.3%自报
ScreenSpot Pro70.3%自报
Hallusion Bench70.0%自报
RefSpatialBench0.68 / 100自报
ERQA60.5%自报
MMLongBench-Doc0.60 / 100自报
SimpleVQA0.56 / 100自报
BabyVision44.6%自报
ODinW41.1%自报
ZEROBench-Sub0.36 / 100自报
SUNRGBD0.35 / 100自报
Nuscene15.2%自报
Hypersim0.13 / 100自报
ZEROBench0.10 / 100自报

AA 评测指数

(Artificial Analysis)
Tau2(Sierra + U Toronto + Vector Institute (2025))
93.9
Gpqa(NYU + Cohere + Anthropic (2023))
85.8
Lcr(Artificial Analysis)
77.7
Ifbench(Google Research (2023))
75.6
Terminalbench Hard(Stanford × Laude Institute (2026))
32.6
Hle(Center for AI Safety + Scale AI (2025))
23.9
Intelligence Index(Artificial Analysis)
22.9

LLM Stats 分类评分

(LLM Stats (zeroeval))
Instruction Following
90
Biology
90
Chat
80
Image To Text
80
Language
80
Legal
80
Math
80
Physics
80
Structured Output
80
Embodied
80
Finance
80
Grounding
80
Chemistry
80
Text-to-image
80
Video
80
Long Context
70
Multimodal
70
Reasoning
70
Spatial Reasoning
70
Frontend Development
70
General
70
Healthcare
70
Economics
70
Vision
70
Search
60
Agents
60
Code
60
Communication
60
Tool Calling
60
Spatial
20
3d
20

定价

输入价格$0.3 / 1M tokens
输出价格$2.4 / 1M tokens
混合价格(3:1)$0.825 / 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: DeepInfra最贵: Alibaba
供应商输入输出
1DeepInfra最便宜
$0
$0
2Novita
$0
$0
3EmpirioLabs AI
$0.086
$0.688
4Alibaba主要
$0.3
$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