Skip to main content
Language modelActive

Qwen 3.6 27B

Alibaba

Released
April 21, 2026
Data date
September 8, 2026
Speed
56.7 tokens/s
Output throughput
Price
$0.3 · $3.2
Input · Output / 1M tokens
Input
Text, Image, Video
Output
Text

Qwen 3.6 27B is an open dense vision-language checkpoint. Qwen highlights frontend workflows and repository-level reasoning as central improvements. Its vision encoder adds image understanding. An option to preserve reasoning context from earlier messages helps with iterative development when several coding steps build on one another.

The model has 27 billion parameters, 262,144 native context tokens, and a documented extension to roughly 1.01 million tokens. Its reasoning trace can be preserved across conversation turns.

Specifications and access

SpecificationValue and source
Weights
Open weightsSource
Access
API, LocalSource
Parameters
27BSource
Context window
262,144 tokensSource
Weights repository
Qwen/Qwen3.6-27BSource
License
Apache-2.0Source
Architecture
hybrid Gated DeltaNet + Gated Attention; dense or sparse MoE with vision encoderSource
Input
Text, Image, VideoSource
Output
TextSource
Reasoning mode
thinking by default; switchable non-thinking/instruct modeSource
Tool calling
YesSource
Structured output
YesSource
Native context window
262,144 tokensSource
Maximum context window
1,010,000 tokensSource
Parameters
27 BSource
Languages
201 languages and dialectsSource
Repository created
2026-04-21 (Repository creation date is a precise publication proxy, but may differ from the public announcement date or an individual checkpoint's first availability.)Source