Passer au contenu principal
AITier

Qwen3.5 122B A10B (Non-reasoning)

AlibabaQwenOpen WeightApache 2.0 · Usage Commercial

Description

Qwen3.5-122B-A10B is a multimodal Mixture-of-Experts model with 122 billion total parameters and 10 billion activated parameters. It combines strong reasoning, coding, long-context, and visual understanding performance with production-friendly efficiency and a native 262K context window.

Date de sortie
2026-02-24
Paramètres
122.0B
Longueur du contexte
262K
Modalités
audio, image, text, video

Radar de capacités

17
general
43
coding
83
reasoning
62
science
60
agents
80
multimodal

Classements

Domaine#RangScoreSource
Capacité agentique47
49.0
LS
Classement codage267
55.0
AA
Classement général245
48.0
AA
Classement multimodal43
59.0
LS
Science228
55.0
AA

Scores de benchmarks (LLM Stats)

(LLM Stats (zeroeval))

Agents

t2-bench79.5%Aut.
VITA-Bench33.6%Aut.
DeepPlanning24.1%Aut.

Chat

IFEvalGoogle Research (2023)93.4%Aut.
Multi-Challenge61.5%Aut.

Code

FullStackBench en62.6%Aut.
FullStackBench zh58.7%Aut.

General

C-Eval91.9%Aut.
MAXIFE87.9%Aut.
Include82.8%Aut.
AndroidWorld_SR66.4%Aut.
NOVA-6358.6%Aut.

Healthcare

MedXpertQA67.3%Aut.
PMC-VQA63.3%Aut.

Instruction Following

IFBench76.1%Aut.

Language

MMLU-Redux94.0%Aut.
MMMLU86.7%Aut.
MMLU-Pro86.7%Aut.
MMLU-ProX82.2%Aut.
WMT24++78.3%Aut.

Long Context

LongBench v260.2%Aut.

Math

HMMT 202591.4%Aut.
HMMT2590.3%Aut.
MathVista-Mini87.4%Aut.
MathVision86.2%Aut.
DynaMath85.9%Aut.
CodeForces0.85 / 3000Aut.
PolyMATH68.9%Aut.

Multimodal

VideoMME w/o sub.83.9%Aut.
MMMU83.9%Aut.
VideoMMMU82.0%Aut.
OSWorld-Verified58.0%Aut.
TIR-Bench53.2%Aut.

Reasoning

Global PIQA88.4%Aut.
GPQANYU + Cohere + Anthropic (2023)86.6%Aut.
LiveCodeBench v678.9%Aut.
CharXiv-R77.2%Aut.
SWE-Bench Verified72.0%Aut.
BrowseComp-zh69.9%Aut.
SuperGPQA67.1%Aut.
AA-LCR66.9%Aut.
BrowseCompOpenAI (2025)63.8%Aut.
Terminal-Bench 2.0Stanford × Laude Institute (2026)49.4%Aut.
Humanity's Last Exam47.5%Aut.
Seal-044.1%Aut.
OJBench39.5%Aut.

Search

WideSearch60.5%Aut.

Tool Calling

BFCL-V472.2%Aut.

Video

MLVU87.3%Aut.

Vision

CountBench0.97 / 100Aut.
VLMsAreBlind96.7%Aut.
AI2D93.3%Aut.
V*93.2%Aut.
MMBench-V1.192.8%Aut.
OCRBench92.1%Aut.
RefCOCO-avg0.91 / 100Aut.
OmniDocBench 1.589.8%Aut.
VideoMME w sub.87.3%Aut.
RealWorldQA85.1%Aut.
EmbSpatialBench0.84 / 100Aut.
MMStar82.9%Aut.
CC-OCR81.8%Aut.
SlakeVQA81.6%Aut.
LingoQA80.8%Aut.
MMMU-Pro76.9%Aut.
MVBench76.6%Aut.
MMVU74.7%Aut.
LVBench74.4%Aut.
ScreenSpot Pro70.4%Aut.
RefSpatialBench0.69 / 100Aut.
Hallusion Bench67.6%Aut.
ERQA62.0%Aut.
SimpleVQA0.62 / 100Aut.
MMLongBench-Doc0.59 / 100Aut.
ODinW44.5%Aut.
BabyVision40.2%Aut.
SUNRGBD0.36 / 100Aut.
ZEROBench-Sub0.36 / 100Aut.
Nuscene15.4%Aut.
Hypersim0.13 / 100Aut.
ZEROBench0.09 / 100Aut.

Indices d'évaluation AA

(Artificial Analysis)
Tau2(Sierra + U Toronto + Vector Institute (2025))
84.5
Gpqa(NYU + Cohere + Anthropic (2023))
82.7
Lcr(Artificial Analysis)
61.3
Ifbench(Google Research (2023))
50.8
Terminalbench V2 1
47.2
Coding Index(Artificial Analysis)
43.3
Terminalbench Hard(Stanford × Laude Institute (2026))
29.5
Intelligence Index(Artificial Analysis)
17.7
Hle(Center for AI Safety + Scale AI (2025))
15.9
Tau Banking
10.3

Scores par catégorie LLM Stats

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

Tarification

Prix d'entrée$0.4 / 1M tokens
Prix de sortie$3.2 / 1M tokens
Prix mixte (3:1)$1.1 / 1M tokens

Vitesse

Tokens/sec143.3
Délai du premier token1.01s
Temps de réponse1.01s

Classement des Prix par Fournisseur

Classement des Prix par Fournisseur

2 fournisseurs

Moins cher: DeepInfraPlus cher: Alibaba
FournisseurEntréeSortie
1DeepInfraMoins cher
$0
$0
2AlibabaPRINCIPAL
$0.4
$3.2

Comparer les prix entre différents fournisseurs API pour ce modèle.

Sources externes

LLM StatsArtificial Analysis