跳轉到主要內容
AITier

Qwen3.5 27B (Reasoning)

AlibabaQwen開源權重Apache 2.0 · 商用許可

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.

發布日期
2026-02-24
參數規模
27.0B
上下文長度
262K
支援模態
audio, image, text, video

能力雷達圖

32
general
40
coding
86
reasoning
57
science
60
agents
80
multimodal

排行榜排名

領域#排名分數來源
智慧體能力模型榜102
38.0
LS
程式碼能力榜123
68.0
AA
通用能力榜86
73.0
AA
多模態榜36
53.0
LS
科學能力115
66.0
AA

基準測試分數 (LLM Stats)

(LLM Stats (zeroeval))

3d

SUNRGBD0.35 / 100自報
Hypersim0.13 / 100自報

Agents

t2-bench79.0%自報
BFCL-V468.5%自報
AndroidWorld_SR64.2%自報
WideSearch61.1%自報
BrowseCompOpenAI (2025)61.0%自報
FullStackBench en60.1%自報
TIR-Bench59.8%自報
FullStackBench zh57.4%自報
OSWorld-Verified56.2%自報
VITA-Bench41.9%自報
Terminal-Bench 2.0Stanford × Laude Institute (2026)41.6%自報
DeepPlanning22.6%自報

Biology

GPQANYU + Cohere + Anthropic (2023)85.5%自報

Chemistry

SuperGPQA65.6%自報

Code

SWE-Bench Verified72.4%自報

Communication

Multi-Challenge60.8%自報

Embodied

EmbSpatialBench0.84 / 100自報

Finance

MMLU-Pro86.1%自報
MMLU-ProX82.2%自報

General

IFEvalGoogle Research (2023)95.0%自報
MMLU-Redux93.2%自報
C-Eval90.5%自報
MAXIFE88.0%自報
Global PIQA87.5%自報
MMMLU85.9%自報
MMMU82.3%自報
Include81.6%自報
MMStar81.0%自報
LiveCodeBench v680.7%自報
IFBench76.5%自報
MMMU-Pro75.0%自報
LongBench v260.6%自報
NOVA-6358.1%自報
SimpleVQA0.56 / 100自報

Grounding

RefCOCO-avg0.91 / 100自報
ScreenSpot Pro70.3%自報
RefSpatialBench0.68 / 100自報

Healthcare

VideoMMMU82.3%自報
SlakeVQA80.0%自報
MedXpertQA62.4%自報
PMC-VQA62.4%自報

Image To Text

OCRBench89.4%自報

Language

LingoQA82.0%自報
WMT24++77.6%自報

Long Context

MLVU85.9%自報
LVBench73.6%自報
AA-LCR66.1%自報
MMLongBench-Doc0.60 / 100自報

Math

HMMT 202592.0%自報
HMMT2589.8%自報
MathVista-Mini87.8%自報
DynaMath87.7%自報
MathVision86.0%自報
CodeForces0.81 / 3000自報
PolyMATH71.2%自報
Humanity's Last Exam48.5%自報

Multimodal

VLMsAreBlind96.9%自報
V*93.7%自報
AI2D92.9%自報
MMBench-V1.192.6%自報
OmniDocBench 1.588.9%自報
VideoMME w sub.87.0%自報
VideoMME w/o sub.82.8%自報
CC-OCR81.0%自報
CharXiv-R79.5%自報
MVBench74.6%自報
MMVU73.3%自報
BabyVision44.6%自報
ZEROBench-Sub0.36 / 100自報
Nuscene15.2%自報
ZEROBench0.10 / 100自報

Reasoning

CountBench0.98 / 100自報
Hallusion Bench70.0%自報
BrowseComp-zh62.1%自報
ERQA60.5%自報
Seal-047.2%自報
OJBench40.1%自報

Spatial Reasoning

RealWorldQA83.7%自報

Vision

ODinW41.1%自報

AA 評測指數

(Artificial Analysis)
Intelligence Index(Artificial Analysis)
34.6
Tau2(Sierra + U Toronto + Vector Institute (2025))
0.9
Gpqa(NYU + Cohere + Anthropic (2023))
0.9
Ifbench(Google Research (2023))
0.8
Lcr(Artificial Analysis)
0.7
Scicode(UIUC + Argonne National Lab (2024))
0.4
Terminalbench Hard(Stanford × Laude Institute (2026))
0.3
Hle(Center for AI Safety + Scale AI (2025))
0.2

LLM Stats 分類評分

(LLM Stats (zeroeval))
Instruction Following
90
Biology
90
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

定價

輸入價格$0.3 / 1M tokens
輸出價格$2.4 / 1M tokens
混合價格(3:1)$0.825 / 1M tokens

速度

Tokens/秒0.0
首Token延遲0.00s
首回答延遲0.00s

供應商價格排行

供應商價格排行

3 個供應商

最便宜: Novita最貴: Alibaba
供應商輸入輸出
1Novita最便宜
$0
$0
2EmpirioLabs AI
$0.086
$0.688
3Alibaba主要
$0.3
$2.4

比較該模型在不同 API 供應商之間的定價。

外部連結

LLM StatsArtificial Analysis