Qwen3.6-35B-A3B

Name: Qwen3.6-35B-A3B
Author: Alibaba

Multimodal

Alibaba

Qwen3.6-35B-A3B is the first open-weight variant of the Qwen3.6 series, a multimodal Mixture-of-Experts model with 35B total parameters and 3B activated.

Key Specifications

Parameters

35.0B

Context

Release Date

April 16, 2026

Average Score

68.9%

API Documentation Repository Model Weights Results Blog

Timeline

Key dates in the model's history

Announcement

April 16, 2026

Last Update

August 29, 2026

Today

October 17, 2026

Technical Specifications

Parameters

35.0B

Training Tokens

Knowledge Cutoff

Family

Capabilities

MultimodalZeroEval

Benchmark Results

Model performance metrics across various tests and benchmarks

Programming

Programming skills tests

SWE-Bench Verified

Internal agent scaffold (bash + file-edit), 200K ctx, temp=1.0 • Self-reported

73.4%

Reasoning

Logical reasoning and analysis

GPQA

• Self-reported

86.0%

Multimodal

Working with images and visual data

AI2D

TEST • Self-reported

92.7%

MMMU

• Self-reported

81.7%

Other Tests

Specialized benchmarks

AIME 2026

Full AIME 2026 (I & II) • Self-reported

92.7%

CC-OCR

• Self-reported

81.9%

C-Eval

• Self-reported

90.0%

CharXiv-R

RQ • Self-reported

78.0%

Claw-Eval

Pass^3 • Self-reported

50.0%

DeepPlanning

• Self-reported

25.9%

EmbSpatialBench

• Self-reported

84.3%

Hallusion Bench

• Self-reported

69.8%

HMMT 2025

February 2025 • Self-reported

90.7%

HMMT25

November 2025 • Self-reported

89.1%

HMMT Feb 26

• Self-reported

83.6%

Humanity's Last Exam

• Self-reported

21.4%

IMO-AnswerBench

• Self-reported

78.9%

LiveCodeBench v6

• Self-reported

80.4%

LVBench

• Self-reported

71.4%

MathVista-Mini

• Self-reported

86.4%

MCP Atlas

Public set, gemini-2.5-pro judge • Self-reported

62.8%

MCP-Mark

GitHub MCP v0.30.3, Playwright responses truncated at 32K • Self-reported

37.0%

MLVU

• Self-reported

86.2%

MMBench-V1.1

EN_V1.1_dev • Self-reported

92.8%

MMLU-Pro

• Self-reported

85.2%

MMLU-Redux

• Self-reported

93.3%

MMMU-Pro

• Self-reported

75.3%

MVBench

• Self-reported

74.6%

NL2Repo

• Self-reported

29.4%

ODinW

13 • Self-reported

50.8%

OmniDocBench 1.5

• Self-reported

89.9%

RealWorldQA

• Self-reported

85.3%

RefCOCO-avg

• Self-reported

92.0%

RefSpatialBench

• Self-reported

64.3%

SimpleVQA

• Self-reported

58.9%

SkillsBench

OpenCode, 78 self-contained tasks, avg of 5 runs • Self-reported

28.7%

SuperGPQA

• Self-reported

64.7%

SWE-bench Multilingual

• Self-reported

67.2%

SWE-Bench Pro

Refined public set • Self-reported

49.5%

TAU3-Bench

Official user model (gpt-5.2 low) + BM25 retrieval • Self-reported

67.2%

Terminal-Bench 2.0

Harbor/Terminus-2, 256K ctx, avg of 5 runs • Self-reported

51.5%

Toolathlon

• Self-reported

26.9%

VideoMME w/o sub.

Without subtitles • Self-reported

82.5%

VideoMME w sub.

With subtitles • Self-reported

86.6%

VideoMMMU

• Self-reported

83.7%

VITA-Bench

claude-4-sonnet judge • Self-reported

35.6%

WideSearch

• Self-reported

60.1%

ZClawBench

Internal real-user-distribution Claw agent benchmark, 256K ctx • Self-reported

52.6%

ZEROBench-Sub

• Self-reported

34.4%

License & Metadata

License

apache-2.0

Announcement Date

April 16, 2026

Last Updated

August 29, 2026

Similar Models

All Models

Qwen3 VL 32B Thinking

Alibaba

MM33.0B

Best score:0.9 (MMLU)

Released:Sep 2025

Qwen3.8-27B

Alibaba

MM27.8B

Best score:0.9 (GPQA)

Released:Aug 2026

Qwen3.6-27B

Alibaba

MM27.8B

Best score:0.9 (GPQA)

Released:Apr 2026

Price:$0.60/1M tokens

Qwen2-VL-72B-Instruct

Alibaba

MM73.4B

Released:Aug 2024

Qwen2.5 VL 72B Instruct

Alibaba

MM72.0B

Released:Jan 2025

QvQ-72B-Preview

Alibaba

MM73.4B

Released:Dec 2024

Qwen2.5 VL 32B Instruct

Alibaba

MM33.5B

Best score:0.9 (HumanEval)

Released:Feb 2025

Qwen3.5-397B-A17B

Alibaba

MM397.0B

Best score:0.9 (GPQA)

Released:Feb 2026

Recommendations are based on similarity of characteristics: developer organization, multimodality, parameter size, and benchmark performance. Choose a model to compare or go to the full catalog to browse all available AI models.