Jackrong/Qwopus3.6-27B-v1-preview

VISIONConcurrency Cost:2Model Size:27BQuant:FP8Ctx Length:32kTool Calling:SupportedPublished:Apr 22, 2026License:apache-2.0Architecture:Transformer0.0K Open Weights Cold

Jackrong/Qwopus3.6-27B-v1-preview is a 27 billion parameter early preview reasoning model built on the Qwen3.6-27B architecture. Developed by Jackrong, it focuses on enhanced reasoning quality, consistent answer structure, and reduced stylistic drift in long-form responses. This model is fine-tuned with a curated dataset emphasizing high-quality reasoning traces, making it suitable for structured reasoning tasks and as a foundation for larger-scale versions. It supports a 32,768 token context length, extensible up to 1,010,000 tokens with YaRN scaling.

Loading preview...

Qwopus3.6-27B-v1-preview: Enhanced Reasoning and Consistency

Jackrong/Qwopus3.6-27B-v1-preview is a 27 billion parameter model, serving as an early preview in the Qwopus series, built upon the Qwen3.6-27B base. This model prioritizes stronger reasoning quality, a more stable answer structure, and reduced stylistic drift in long-form responses. It achieves this through a refined supervised fine-tuning approach, utilizing a cleaned dataset primarily from Kassadin88/Claude-Distillation-Dataset and other reasoning-focused sources.

Key Capabilities

  • Structured Reasoning: Designed for tasks requiring deliberate and coherent thought processes.
  • Consistent Output Style: Maintains a uniform answer style across various tasks, reducing inconsistencies.
  • Cross-Source Distillation Alignment: Improves alignment when distilling knowledge from diverse sources.
  • Long Context Support: Natively handles up to 32,768 tokens, extensible to 1,010,000 tokens using YaRN scaling techniques.
  • Multimodal (Base Model): Inherits vision and video understanding capabilities from its Qwen3.6-27B base, including agentic coding and thinking preservation.

Good For

  • Applications requiring reliable and structured reasoning outputs.
  • Developers seeking a model with predictable response styles for integration into workflows.
  • Use cases benefiting from extended context windows for complex problems.
  • As a strong foundation for further fine-tuning or larger-scale model development.

Finetunes
2 models