RedTeaDev/Qwen-3.5-9b-gemini-3.1-pro-Claude-Opus-4.6-high-reasoning is a reasoning model fine-tuned on top of the Qwen3.5 architecture. This model aims to distill the Chain-Of-Thought from Claude-4.6 Opus and Gemini-3.1-Pro (High Reasoning).

By default this model has thinking=False, if you are using GGUF and wanted to enable thinking, you can add {%- set enable_thinking = true %} to your chat template

The model aims to address Qwen3.5 model family's tendency toward excessive reasoning toward simple user query, Through distillation and structural imitation of Claude 4.6 Opus and Gemini 3.1 Pro, this fine-tuned model adopts a more efficient thinking pattern.

Compared to v1 version of the fine-tuned model, this model has also include the zake7749/OpenScience-Chinese-Reasoning-SFT dataset to preserve the ability for chinese instruction following

This qwen3_5 model was trained 2x faster with Unsloth and Huggingface's TRL library.