Home
Aliyun Tongyi Qianwen again open-sources Qwen3.6-27B, showing strong coding in a compact model.
On April 22, Alibaba Cloud's Tongyi Qianwen team announced a major update to its open-source lineup, officially releasing Qwen3.6-27B, a dense multimodal model with 27 billion parameters. As the most requested model configuration among developers, this version not only expands the Qwen series product matrix but also represents a deep evolution in intelligent agent programming and multimodal reasoning, while preserving the advantages of a dense architecture.

Performance Leap: Programming Capabilities Outperform 15x Larger MoE Models
In this release, the standout achievement is its remarkable performance. Despite having only 27 billion parameters, Qwen3.6-27B surpasses the previous version, Qwen3.5-397B-A17B (with 397 billion total parameters), across various programming benchmarks. Data shows it scored 77.2 on the SWE-bench Verified test, which evaluates code repair skills, and showed even greater improvements on tasks like SkillsBench. This means developers can access a top-tier programming assistant without needing complex MoE (Mixture of Experts) routing, greatly reducing deployment barriers.

All-Round Multimodal: Supports Mixed Image and Video Input
Beyond its strong logical reasoning, Qwen3.6-27B also excels in the visual language domain. It natively supports multimodal processing, seamlessly parsing mixed inputs of images, videos, and text—covering use cases like visual reasoning, deep document understanding, and interactive visual question answering. According to the official announcement, its multimodal capabilities match those of the higher-parameter Qwen3.6-35B-A3B, ensuring high-precision outputs for multimodal tasks.

Ecosystem Integration: Deeply Aligns with Developers' Mainstream Workflows
To accelerate the transition from technology to productivity, the model's open-source weights have been released simultaneously on Hugging Face and ModelScope (Moba) communities, supporting local deployment. Additionally, the Alibaba Cloud Bailian platform will soon offer API access, and notably retains the "preserve_thinking" feature to enable full traceability of the thought chain in agent tasks. Currently, Qwen3.6-27B has been seamlessly integrated with mainstream coding assistants like Claude Code and Qwen Code, aiming to deliver more accurate and context-aware coding support to developers worldwide.
Related article
ByteDance’s Seed launches global campus drive, offering virtual shares to win top large model talent
In the competitive landscape of large language models, securing top-tier talent remains the most critical strategic asset.On April 1st, ByteDance announced the launch of its Seed global campus recruitment initiative, part of its large model talent de
Suno to Watermark Songs Amid Legal Battles
Suno, the platform enabling users to generate AI-created music, has unveiled new features to label platform-produced tracks, restrict downloads, and update community standards to curb unauthorized replicas. These updates arrive as Suno confronts mult
Musk Admits Grok Build Leaked User Code, Promises to Erase All Historical Data
Elon Musk directly addressed the privacy controversy surrounding Grok Build, beginning with a simple "True" to confirm the incident's validity. He pledged that all user data previously uploaded to SpaceXAI would be permanently erased, stating, "not a
Related Special Topic Recommendations
Comments (0)
0/500
On April 22, Alibaba Cloud's Tongyi Qianwen team announced a major update to its open-source lineup, officially releasing Qwen3.6-27B, a dense multimodal model with 27 billion parameters. As the most requested model configuration among developers, this version not only expands the Qwen series product matrix but also represents a deep evolution in intelligent agent programming and multimodal reasoning, while preserving the advantages of a dense architecture.

Performance Leap: Programming Capabilities Outperform 15x Larger MoE Models
In this release, the standout achievement is its remarkable performance. Despite having only 27 billion parameters, Qwen3.6-27B surpasses the previous version, Qwen3.5-397B-A17B (with 397 billion total parameters), across various programming benchmarks. Data shows it scored 77.2 on the SWE-bench Verified test, which evaluates code repair skills, and showed even greater improvements on tasks like SkillsBench. This means developers can access a top-tier programming assistant without needing complex MoE (Mixture of Experts) routing, greatly reducing deployment barriers.

All-Round Multimodal: Supports Mixed Image and Video Input
Beyond its strong logical reasoning, Qwen3.6-27B also excels in the visual language domain. It natively supports multimodal processing, seamlessly parsing mixed inputs of images, videos, and text—covering use cases like visual reasoning, deep document understanding, and interactive visual question answering. According to the official announcement, its multimodal capabilities match those of the higher-parameter Qwen3.6-35B-A3B, ensuring high-precision outputs for multimodal tasks.

Ecosystem Integration: Deeply Aligns with Developers' Mainstream Workflows
To accelerate the transition from technology to productivity, the model's open-source weights have been released simultaneously on Hugging Face and ModelScope (Moba) communities, supporting local deployment. Additionally, the Alibaba Cloud Bailian platform will soon offer API access, and notably retains the "preserve_thinking" feature to enable full traceability of the thought chain in agent tasks. Currently, Qwen3.6-27B has been seamlessly integrated with mainstream coding assistants like Claude Code and Qwen Code, aiming to deliver more accurate and context-aware coding support to developers worldwide.
ByteDance’s Seed launches global campus drive, offering virtual shares to win top large model talent
In the competitive landscape of large language models, securing top-tier talent remains the most critical strategic asset.On April 1st, ByteDance announced the launch of its Seed global campus recruitment initiative, part of its large model talent de
Suno to Watermark Songs Amid Legal Battles
Suno, the platform enabling users to generate AI-created music, has unveiled new features to label platform-produced tracks, restrict downloads, and update community standards to curb unauthorized replicas. These updates arrive as Suno confronts mult
Musk Admits Grok Build Leaked User Code, Promises to Erase All Historical Data
Elon Musk directly addressed the privacy controversy surrounding Grok Build, beginning with a simple "True" to confirm the incident's validity. He pledged that all user data previously uploaded to SpaceXAI would be permanently erased, stating, "not a











