DavidAU/Qwen3.5-27B-Claude-4.6-OS-INSTRUCT
DavidAU/Qwen3.5-27B-Claude-4.6-OS-INSTRUCT is a 27 billion parameter instruction-tuned causal language model, fine-tuned by DavidAU using Unsloth on the Qwen 3.5 dense model with the Claude-4.6-OS dataset. This model features altered reasoning/thinking blocks for improved performance and is vision-capable, demonstrating strong benchmarks across various language and multimodal tasks. It excels in complex reasoning and general agent capabilities, supporting a native context length of 262,144 tokens.
Loading preview...
Model Overview
DavidAU/Qwen3.5-27B-Claude-4.6-OS-INSTRUCT is a 27 billion parameter multimodal language model, fine-tuned by DavidAU from the Qwen 3.5 base model using the Claude-4.6-OS dataset. This fine-tuning process, conducted via Unsloth, focused on optimizing the model's reasoning and thinking mechanisms, resulting in improved performance without negatively impacting its strong foundational benchmarks. The model supports a native context length of 262,144 tokens, extensible up to 1,010,000 tokens using YaRN scaling techniques.
Key Capabilities
- Enhanced Reasoning: Features altered reasoning/thinking blocks, leading to Gemini Pro-like reasoning in many cases, and a mix of Gemini/Qwen depending on task complexity.
- Multimodal Understanding: Fully vision-capable, tested and working with new training, and supports video input for comprehensive understanding.
- Strong Benchmarks: Demonstrates competitive performance across a wide range of benchmarks, including MMLU-Pro (86.1%), IFEval (95.0%), MMMU (82.3%), and MathVision (86.0%).
- Agentic Usage: Excels in tool calling capabilities, with recommendations for use with Qwen-Agent and Qwen Code for building agent applications.
- Long Context Handling: Natively supports 262,144 tokens and can be extended to over 1 million tokens using YaRN, making it suitable for ultra-long text processing.
Good for
- Complex Reasoning Tasks: Its optimized reasoning blocks make it suitable for tasks requiring deep thought and problem-solving.
- Multimodal Applications: Ideal for applications involving image and video understanding, such as visual question answering and content analysis.
- Agent Development: Recommended for building AI agents that require robust tool-calling and interaction capabilities.
- Long Document Analysis: Its extensive context window makes it effective for processing and understanding very long texts and documents.