longsesh/Qwen3.8-27B

Hugging Face
VISIONPricing:Input $1.6 / Cached $0.15 / Output $12Concurrent Unit Cost:2Model Size:27BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Sep 27, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Warm

Qwen3.8-27B is a 27 billion parameter causal language model developed by Qwen, built on the Qwen3.5 architectural foundation. This model integrates native vision-language understanding for images and videos, and features flexible thinking control for complex, multi-step agentic tasks. It delivers substantial gains across coding, professional work, research, and long-horizon agentic applications, with a native context length of 262,144 tokens, extensible up to 1,000,000 tokens.

Loading preview...

Qwen3.8-27B: Advanced Agentic and Multimodal Capabilities

Qwen3.8-27B, developed by Qwen, represents the latest generation in the Qwen open-model family, building upon the Qwen3.5 architecture. This 27 billion parameter causal language model is designed for enhanced performance across a range of demanding applications, including coding, professional tasks, research, and long-horizon agentic workflows.

Key Capabilities and Enhancements

  • Comprehensive Core Improvements: Significant gains in coding proficiency, professional task execution, research assistance, and long-horizon agentic task completion.
  • Robust Agent Execution: Features stronger autonomous planning and improved handling of environmental feedback, leading to more reliable end-to-end task completion.
  • Flexible Thinking Control: Offers a default thinking mode that can be disabled per request, with tunable reasoning depth via reasoning_effort and retention of historical reasoning context through preserve_thinking.
  • Native Vision-Language Understanding: Provides integrated support for understanding both images and videos, ranging from STEM diagrams and documents to hour-scale video content.
  • Extended Context Length: Natively supports up to 262,144 tokens, with extensibility up to 1,000,000 tokens for ultra-long texts using techniques like RoPE scaling.
  • Downstream Compatibility: Designed for broader support with popular harnesses and development tools, facilitating easier integration into existing tech stacks.

Recommended Use Cases

  • Complex Agentic Workflows: Ideal for tasks requiring autonomous planning, multi-step execution, and reliable completion, benefiting from its enhanced agent execution and flexible thinking control.
  • Multimodal Applications: Suited for scenarios involving the interpretation of visual data, including image analysis, video understanding, and processing documents with embedded diagrams.
  • Coding and Research: Offers improved performance for coding tasks and research-oriented applications, leveraging its comprehensive core capability enhancements.
  • Long-Context Processing: Effective for applications requiring the processing of very long documents or conversations, thanks to its extended context window.

Finetunes
1 models