Qwen3.6-27B

Name: Qwen3.6-27B
Author: Alibaba

Multimodal

Alibaba

Qwen3.6-27B is a 27.8B-parameter open-weight multimodal model from the Qwen team, released in April 2026. It delivers strong visual counting and reasoning, topping CountBench and VLMsAreBlind, and is Apache 2.0 licensed. It is served via Novita.

Key Specifications

Parameters

27.8B

Context

262.1K

Release Date

April 21, 2026

Average Score

77.5%

API Documentation Research Paper Results Blog

Timeline

Key dates in the model's history

Announcement

April 21, 2026

Last Update

August 27, 2026

Today

September 10, 2026

Technical Specifications

Parameters

27.8B

Training Tokens

Knowledge Cutoff

Family

Capabilities

MultimodalZeroEval

Pricing & Availability

Input (per 1M tokens)

$0.60

Output (per 1M tokens)

$3.60

Max Input Tokens

262.1K

Max Output Tokens

65.5K

Supported Features

Function CallingStructured OutputCode ExecutionWeb SearchBatch InferenceFine-tuning

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, top_p=0.95 • Self-reported

77.2%

Reasoning

Logical reasoning and analysis

GPQA

• Self-reported

87.8%

Multimodal

Working with images and visual data

MMMU

• Self-reported

82.9%

Other Tests

Specialized benchmarks

CountBench

Self-reported by the model provider • Self-reported

98.0%

VLMsAreBlind

Self-reported by the model provider • Self-reported

97.0%

Self-reported by the model provider • Self-reported

95.0%

AIME 2026

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

94.1%

AndroidWorld

• Self-reported

70.3%

C-Eval

• Self-reported

91.4%

CC-OCR

• Self-reported

81.2%

CharXiv-R

RQ • Self-reported

78.4%

Claw-Eval

Pass^3 • Self-reported

60.6%

DynaMath

• Self-reported

85.6%

EmbSpatialBench

• Self-reported

84.6%

ERQA

• Self-reported

62.5%

HMMT 2025

February 2025 • Self-reported

93.8%

HMMT Feb 26

• Self-reported

84.3%

HMMT25

November 2025 • Self-reported

90.7%

Humanity's Last Exam

• Self-reported

24.0%

IMO-AnswerBench

• Self-reported

80.8%

LiveCodeBench v6

• Self-reported

83.9%

MathVista-Mini

• Self-reported

87.4%

MLVU

• Self-reported

86.6%

MMBench-V1.1

EN_V1.1_dev • Self-reported

92.3%

MMLU-Pro

• Self-reported

86.2%

MMLU-Redux

• Self-reported

93.5%

MMMU-Pro

• Self-reported

75.8%

MMStar

• Self-reported

81.4%

MVBench

• Self-reported

75.5%

NL2Repo

• Self-reported

36.2%

OCRBench

• Self-reported

89.4%

QwenWebBench

Internal front-end code generation benchmark, BT/Elo rating • Self-reported

74.4%

RealWorldQA

• Self-reported

84.1%

RefCOCO-avg

• Self-reported

92.5%

RefSpatialBench

• Self-reported

70.0%

SimpleVQA

• Self-reported

56.1%

SkillsBench

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

48.2%

SuperGPQA

• Self-reported

66.0%

SWE-bench Multilingual

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

71.3%

SWE-Bench Pro

Refined public set (problematic tasks corrected). Internal agent scaffold, 200K ctx • Self-reported

53.5%

Terminal-Bench 2.0

Harbor/Terminus-2 harness, 256K ctx, max_tokens=80K, avg of 5 runs • Self-reported

59.3%

VideoMME w sub.

With subtitles • Self-reported

87.7%

VideoMMMU

• Self-reported

84.4%

ZClawBench

Real-user-distribution Claw agent benchmark, 256K ctx, temp=0.6 • Self-reported

53.4%

License & Metadata

License

apache-2.0

Announcement Date

April 21, 2026

Last Updated

August 27, 2026

Similar Models

All Models

Qwen3.6-35B-A3B

Alibaba

MM35.0B

Best score:0.9 (GPQA)

Released:Apr 2026

Qwen3.8-27B

Alibaba

MM27.8B

Best score:0.9 (GPQA)

Released:Aug 2026

Qwen3 VL 32B Thinking

Alibaba

MM33.0B

Best score:0.9 (MMLU)

Released:Sep 2025

Qwen2.5 VL 72B Instruct

Alibaba

MM72.0B

Released:Jan 2025

QvQ-72B-Preview

Alibaba

MM73.4B

Released:Dec 2024

Qwen2-VL-72B-Instruct

Alibaba

MM73.4B

Released:Aug 2024

Qwen2.5 VL 32B Instruct

Alibaba

MM33.5B

Best score:0.9 (HumanEval)

Released:Feb 2025

Qwen3.6 Plus

Alibaba

Best score:0.9 (GPQA)

Released:Mar 2026

Price:$0.50/1M tokens

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.