khazarai/Qwen3.8-27B-Fable-5-Coding-Distilled

VISIONPricing:Input $1.6 / Cached $0.15 / Output $12Concurrent Unit Cost:2Model Size:27BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Sep 15, 2026License:apache-2.0Architecture:Transformer0.0K Open Weights Featherless Exclusive Cold

khazarai/Qwen3.8-27B-Fable-5-Coding-Distilled is a 27 billion parameter language model, fine-tuned from Qwen/Qwen3.8-27B, specifically optimized for coding, debugging, and agentic reasoning tasks. It was trained on 195 million tokens of distilled reasoning traces to promote grounded, tool-using behavior. This model excels at inspecting project state and providing evidence-based solutions for software engineering problems, making it ideal for agentic coding assistants and root-cause analysis.

Loading preview...

Model Overview

khazarai/Qwen3.8-27B-Fable-5-Coding-Distilled is a specialized fine-tune of the Qwen/Qwen3.8-27B base model, focusing on enhancing its capabilities in coding, debugging, and agentic reasoning. This 27 billion parameter model was trained on 195 million tokens of distilled reasoning traces, specifically designed to foster "check before you fix" behavior rather than speculative problem-solving.

Key Capabilities & Differentiators

  • Grounded Reasoning: Emphasizes inspecting actual project files, test harnesses, and CI configurations to verify assumptions before proposing solutions.
  • Agentic Tool-Use: Promotes interaction with real-world state, such as reading files and working within existing code constraints.
  • Debugging Focus: Optimized for tasks like identifying root causes of CI failures, lint errors, and shell script bugs.
  • Concise & Evidence-Based: Aims for short, grounded responses based on verified information, avoiding lengthy, speculative chains of thought.

Intended Use Cases

  • Agentic Coding Assistants: Ideal for systems that need to inspect a repository's state before taking action.
  • Debugging & Root-Cause Analysis: Effective for diagnosing issues in CI pipelines, shell scripts, and general code.
  • Code Review & Refactoring: Suitable for tasks requiring adherence to existing project constraints.

Limitations

This model inherits the limitations of its Qwen3.8-27B base, including potential hallucination on unfamiliar codebases. Its distillation was specifically for coding and agentic debugging, so its performance on general chat or non-coding reasoning tasks has not been separately evaluated.

Finetunes
2 models