Qwen 3.8 27B

All projects

Apps & Tools

A powerful 27B native vision-language model with flexible thinking control.

byAhmad

Playable

Qwen 3.8 27B media is blocked

Allow external media to connect to the provider and play this content.

Description

Qwen3.8-27B is a 27-billion parameter dense model optimized for coding, research, and long-horizon agentic tasks. It features native vision-language understanding for image and video analysis and includes a 'thinking mode' that allows for tunable reasoning depth during complex task execution.

Descriptions, tags, and model credits may be AI-generated or inferred from public sources and can be incomplete or wrong. Learn more.

Details

Why this is interesting: Major open model release introducing native vision-language support and flexible 'thinking mode' reasoning.
Attributed to: @theahmadosman
Built with: Grok 4.6 Opus 5 Opus 4.8 Qwen 3.7 Max Opus 4.7 Opus 4.6 Qwen3 8B Qwen3.8-Max Qwen3.8-27B Qwen3 14B Qwen3 30B-A3B Muse Glimmer Qwen3.7 Flash Qwen3.7 Plus Qwen3.6 Opus 4.5
Links: huggingface.co
View tweet
Tags: #llm #vision-language #open-weights #reasoning

Discovered: X
First posted: X · Aug 14, 2026
Listed: Added 12d ago
Build: Strong
Models: Strong

Qwen 3.8 27B

Description

Timeline

Qwen 3.8 27B

Description

Timeline