Qwen3.5 35B A3B (Non-reasoning)

AlibabaQwenOpen WeightApache 2.0 · Commercial OK

Description

Qwen3.5-35B-A3B is a multimodal Mixture-of-Experts model with 35 billion total parameters and 3 billion activated parameters. It combines strong reasoning, coding, agentic, and visual understanding performance with production-friendly efficiency and a native 262K context window.

Release Date

2026-02-24

Parameters

35.0B

Context Length

262K

Modalities

image, text, video

Capability Radar

general

coding

reasoning

scienceest.

agents

multimodal

Science uses a reasoning proxy when dedicated science benchmarks are unavailable.

Rankings

Domain	#Rank	Score	Source
Agents & Tools	53	56.0	LS
Code Ranking	227	37.0	AA
General Ranking	157	57.0	AA
Multimodal Ranking	62	69.0	LS
Reasoning	52	68.0	LS
Science	144	54.0	AA

Benchmark Scores (LLM Stats)

3d

SUNRGBD

0.33 / 100SR

Hypersim

0.13 / 100SR

Agents

t2-bench

81.2%SR

AndroidWorld_SR

71.1%SR

BFCL-V4

67.3%SR

BrowseComp

61.0%SR

FullStackBench en

58.1%SR

WideSearch

57.1%SR

TIR-Bench

55.5%SR

FullStackBench zh

55.0%SR

OSWorld-Verified

54.5%SR

Terminal-Bench 2.0

40.5%SR

VITA-Bench

31.9%SR

DeepPlanning

22.8%SR

Biology

GPQA

84.2%SR

Chemistry

SuperGPQA

63.4%SR

Code

SWE-Bench Verified

69.2%SR

Communication

Multi-Challenge

60.0%SR

Embodied

EmbSpatialBench

0.83 / 100SR

Finance

MMLU-Pro

85.3%SR

MMLU-ProX

81.0%SR

General

MMLU-Redux

93.3%SR

IFEval

91.9%SR

C-Eval

90.2%SR

MAXIFE

86.6%SR

Global PIQA

86.6%SR

MMMLU

85.2%SR

MMStar

81.9%SR

MMMU

81.4%SR

Include

79.7%SR

MMMU-Pro

75.1%SR

LiveCodeBench v6

74.6%SR

IFBench

70.2%SR

LongBench v2

59.0%SR

SimpleVQA

0.58 / 100SR

NOVA-63

57.1%SR

Grounding

RefCOCO-avg

0.89 / 100SR

ScreenSpot Pro

68.6%SR

RefSpatialBench

0.64 / 100SR

Healthcare

VideoMMMU

80.4%SR

SlakeVQA

78.7%SR

PMC-VQA

62.0%SR

MedXpertQA

61.4%SR

Image To Text

OCRBench

91.0%SR

Language

LingoQA

79.2%SR

WMT24++

76.3%SR

Long Context

MLVU

85.6%SR

LVBench

71.4%SR

MMLongBench-Doc

0.59 / 100SR

AA-LCR

58.5%SR

Math

HMMT25

89.2%SR

HMMT 2025

89.0%SR

MathVista-Mini

86.2%SR

DynaMath

85.0%SR

MathVision

83.9%SR

CodeForces

0.82 / 3000SR

PolyMATH

64.4%SR

Humanity's Last Exam

47.4%SR

Multimodal

VLMsAreBlind

97.0%SR

92.7%SR

AI2D

92.6%SR

MMBench-V1.1

91.5%SR

OmniDocBench 1.5

89.3%SR

VideoMME w sub.

86.6%SR

VideoMME w/o sub.

82.5%SR

CC-OCR

80.7%SR

CharXiv-R

77.5%SR

MVBench

74.8%SR

MMVU

72.3%SR

BabyVision

38.4%SR

ZEROBench-Sub

0.34 / 100SR

Nuscene

14.6%SR

ZEROBench

0.08 / 100SR

Reasoning

CountBench

0.98 / 100SR

BrowseComp-zh

69.5%SR

Hallusion Bench

67.9%SR

ERQA

64.8%SR

Seal-0

41.4%SR

OJBench

36.0%SR

Spatial Reasoning

RealWorldQA

84.1%SR

Vision

ODinW

42.6%SR

AA Evaluation Indices

Intelligence Index

30.7

Coding Index

16.8

Tau2

0.9

Gpqa

0.8

Lcr

0.6

Ifbench

0.4

Scicode

0.3

Hle

0.1

Terminalbench Hard

0.1

LLM Stats Category Scores

Structured Output

Text-to-image

Video

Biology

Embodied

Finance

General

Image To Text

Instruction Following

Language

Legal

Math

Physics

Spatial Reasoning

Vision

Chemistry

Frontend Development

Grounding

Healthcare

Long Context

Multimodal

Reasoning

Tool Calling

Code

Communication

Economics

Agents

Spatial

Pricing

Input Price$0.25 / 1M tokens

Output Price$2 / 1M tokens

Blended Price (3:1)$0.688 / 1M tokens

Speed

Tokens/sec138.7 tokens/s

Time to First Token1.15s

Time to Answer1.15s

Available Providers

(LS internal units)

Provider	Input Price	Output Price
Novita	250K	2.0M

External Sources

LLM Stats