clzoro/Qwen3.5-4B-KIMI-Distill

VISIONPricing:Input $0.4 / Cached $0.08 / Output $0.8Concurrent Unit Cost:1Model Size:4.5BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Apr 9, 2026License:apache-2.0Architecture:Transformer0.0K Open Weights Featherless Exclusive Cold

clzoro/Qwen3.5-4B-KIMI-Distill is a 4.5 billion parameter causal language model developed by Kassadin88, distilled from KIMI-K2.5 and fine-tuned from Qwen3.5-4B. It is specifically enhanced for reasoning tasks across domains like coding, science, and mathematics, trained on 554K high-quality reasoning traces totaling 2 billion tokens. This model also inherits vision-language capabilities from its base model and supports a 32K context length, making it suitable for complex problem-solving requiring detailed step-by-step logic.

Loading preview...

Qwen3.5-4B-KIMI-Distill: Reasoning-Enhanced Language Model

This model, developed by Kassadin88, is a 4.5 billion parameter language model fine-tuned from Qwen3.5-4B. Its core distinction lies in its reasoning enhancement, achieved through distillation from the powerful KIMI-K2.5 model. It was trained on an extensive dataset of 554,381 high-quality chain-of-thought reasoning samples, comprising approximately 2 billion tokens.

Key Capabilities

  • Enhanced Reasoning: Specialized in generating detailed, step-by-step reasoning traces across various complex domains.
  • Multi-Domain Expertise: Strong performance in coding (60% of training data), science (15%), mathematics (10%), computer science (5%), and logical reasoning (5%).
  • Vision-Language Integration: Inherits multimodal capabilities from its Qwen3.5-4B base, allowing for processing of both text and visual inputs.
  • Long Context Window: Supports a context length of 32,768 tokens, beneficial for intricate problems and extended dialogues.

Good For

  • Complex Problem Solving: Ideal for tasks requiring logical deduction, mathematical calculations, and scientific explanations.
  • Code Generation and Analysis: Excels in generating well-structured code and providing explanations for programming challenges.
  • Educational Applications: Useful for generating detailed solutions and explanations for academic problems.
  • Research in AI Reasoning: A valuable tool for exploring and developing advanced reasoning capabilities in smaller models.

Finetunes
1 models