Skip to content

Qwen3.8 27B

Qwen/Qwen3.8-27B
Open Source · chat · open-weights
Open Alert me on changes
Context
262.1K
Max output
Weights
Open
API $/1M
$0.45 / $3.20
Modalities
text · image
Released
05 Aug 2026
License: apache-2.0 · Qwen/Qwen3.8-27B
Download image Share on X Share on LinkedIn
AI summary
● machine-written

Qwen3.8 27B: Open-weights 27B vision-language model from Alibaba

Qwen3.8 27B is a 27-billion-parameter open-weights dense model from the Qwen Team at Alibaba that accepts text and image inputs. It is designed for coding, professional workflows, research, multimodal interaction, and long-running agent tasks, with flexible thinking capability. The model supports a 262K token context window and was released in August 2026.

What's new
  • Open-weights model with Apache 2.0 license
  • Supports text and image modalities
  • Flexible thinking mode that can be enabled or disabled
  • 262K token context window
Best for
Coding and professional workflowsMultimodal agent tasksResearch and analysisLong-running autonomous tasks
Sources

Source: https://huggingface.co/Qwen/Qwen3.8-27B