DataLearner logo
QW

Qwen3.5-27B

Reasoning modelQwen3.5

Qwen3.5-27B

Release date: 2026-02-25Updated: 2026-07-17 21:58:07.3382,552
Live demoGitHubHugging FaceCompare
Parameters
27B
Context length
1010K
Chinese support
Supported
Reasoning ability

Qwen3.5-27B is a reasoning model from Alibaba, released on 2026-02-25. It accepts text, image, and video input and returns text output. The cataloged parameter count is 27B. The recorded context window is 1010K, and the recorded maximum output is 248320. Cataloged capabilities include Reasoning model and Multilingual. The checkpoint is listed under the Qwen License, with the license link included in the model references. Use the linked references to confirm current access, licensing, and provider-specific limits.

Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology

Qwen3.5-27B

Model basics

Reasoning traces
Supported
Thinking modes
Thinking Level · Extended (Default)Standard Mode
Context length
1010K tokens
Max output length
248320 tokens
Model type
Reasoning model
Modality (in / out)
Text, Image, Video → Text
Release date
2026-02-25
Model file size
55.6 GB
MoE architecture
No
Total params / Active params
27B / N/A
Knowledge cutoff
No data
Qwen3.5-27B

Open source & experience

Code license
Qwen License
Weights license
Qwen License- Commercial use permitted
GitHub repo
https://github.com/QwenLM/Qwen3.5
Hugging Face
https://huggingface.co/Qwen/Qwen3.5-27B
Live demo
https://chat.qwen.ai/
Qwen3.5-27B

Official resources

Paper
Paper
DataLearnerAI blog
No blog post yet
Qwen3.5-27B

API details

API speed
4/5
No public API pricing yet.
Qwen3.5-27B

Benchmark Results

Qwen3.5-27B currently shows benchmark results led by Pinch Bench (2 / 37, score 90), IF Bench (4 / 30, score 76.50), MMLU Pro (19 / 132, score 86.10). This page also consolidates core specs, context limits, and API pricing so you can evaluate the model from benchmark results and deployment constraints together.

Thinking
Tool usage
Internet

General Knowledge

5 evaluations
Benchmark / mode
Score
Rank/total
C-Eval
Thinking Mode
90.50
6 / 10
MMLU Pro
Thinking Mode
86.10
19 / 132
GPQA Diamond
Thinking Mode
85.50
52 / 187
HLE
Thinking Mode
24.30
103 / 170
HLE
Thinking ModeTools
48.50
31 / 170

Coding and Software Engineer

3 evaluations
Benchmark / mode
Score
Rank/total
CodeForces
Thinking Mode
1899
15 / 16
LiveCodeBench
Thinking ModeTools
80.70
27 / 123
SWE-bench Verified
Thinking Mode
72.40
52 / 111

Multimodal Understanding

2 evaluations
Benchmark / mode
Score
Rank/total
MMMU
Thinking Mode
82.30
9 / 29
SimpleVQA
Thinking Mode
56
3 / 3

Agent Level Benchmark

1 evaluations
Benchmark / mode
Score
Rank/total
τ²-Bench
Thinking ModeTools
79
17 / 43

Instruction Following

1 evaluations
Benchmark / mode
Score
Rank/total
IF Bench
Thinking Mode
76.50
4 / 30

AI Agent - Information Search

2 evaluations
Benchmark / mode
Score
Rank/total
BrowseComp
Thinking ModeToolsInternet
61
33 / 52
BrowseComp
Thinking ModeTools
61
33 / 52

AI Agent - Tool Usage

2 evaluations
Benchmark / mode
Score
Rank/total
OSWorld-Verified
Thinking ModeTools
56.20
17 / 20
Terminal Bench 2.0
Thinking ModeTools
41.60
43 / 47

Long Context

2 evaluations
Benchmark / mode
Score
Rank/total
AA-LCR
Thinking Mode
66.10
8 / 14
LongBench v2
Standard Mode
60.60
7 / 11

Claw-style Agent Evaluation

2 evaluations
Benchmark / mode
Score
Rank/total
Pinch Bench
Thinking ModeTools
90
2 / 37
Claw Bench
Thinking ModeTools
75.20
26 / 29
View benchmark analysisCompare with other models

Compare with other models

Qwen3.5-27B

Publisher

Qwen3.5-27B

Model Overview

Qwen3.5-27B is a reasoning model from Alibaba, released on 2026-02-25.

It accepts text, image, and video input and produces text output. Its cataloged capabilities include Reasoning model and Multilingual. The cataloged parameter count is 27B. The recorded context window is 1010K, and the recorded maximum output is 248320.

The checkpoint is listed under the Qwen License, with the license link included in the model references. The evaluation section contains 20 cataloged benchmark results with their recorded modes and scores. The page links 4 release, model-card, repository, or provider references for checking the underlying claims. Specifications, availability, and prices can change; undisclosed values are intentionally left unstated.

Qwen3.5-27B

FAQ

What is Qwen3.5-27B?

Qwen3.5-27B is a reasoning model from Alibaba, released on 2026-02-25. It accepts text, image, and video input and returns text output. The cataloged parameter count is 27B. The recorded context window is 1010K, and the recorded maximum output is 248320. Cataloged capabilities include Reasoning model and Multilingual. The checkpoint is listed under the Qwen License, with the license link included in the model references. Use the linked references to confirm current access, licensing, and provider-specific limits.

What input and output modalities does Qwen3.5-27B support?

The current model record lists text, image, and video as input and text as output.

What are the main recorded specifications for Qwen3.5-27B?

The cataloged parameter count is 27B. The recorded context window is 1010K, and the recorded maximum output is 248320. Fields without a source-backed value remain undisclosed.

Why is no API price shown for Qwen3.5-27B?

Open-weight checkpoint; no first-party hosted price published for this exact checkpoint.

Are benchmark results available for Qwen3.5-27B?

The evaluation section contains 20 cataloged benchmark results with their recorded modes and scores. Compare only results that use the same benchmark version and evaluation mode.

Is Qwen3.5-27B open source?

The checkpoint is listed under the Qwen License, with the license link included in the model references. Review the linked license text before commercial or derivative use.

DataLearner on WeChat

Follow DataLearner on WeChat for AI model updates and research notes.

DataLearner WeChat QR code