adityabhushannagar/Qwen-3.5-9B-Gemini-Opus-merge-GGUF

Qwen3.5-9B-Gemini-Opus-merge

This model was made for learning the aspect of merging two already finetuned models for understanding the aspect of model-merging and how any how it may not work. This model doesnt intends to be used very competitvely as an alternative to the other models. Your mileage with this model can vary!

Intention:

I wanted to learn how to create a custom merge of two different finetunes of a single model... Since at the time this model has been created MergeKit doesnt natively supports Qwen3.5 for merging or any other action. This model was created manually with proper support from claude to create the scafolding the merge.

Warning:

This model is not inteneded for any usage in production nor in any environment. This model was created solely for the purpose of learning how model merging can be used for creation of custom models on local hardware.

Hardware used:

This model was merged locally on a RTX 3050 Mobile laptop with 16GB DDR5 Ram laptop with 6GB VRam available in the model.

Safetensors Can be found at:

https://huggingface.co/adityabhushannagar/Qwen-3.5-9B-Gemini-Opus-merge/

Recommended:

• Use Q4_0 and Q4_1 for general purposes
• Q6_K gives almost the same result as the Q8_0 but has a very slight quality drop
• Q8_0 provides almost the same quality as F16 but takes much lower space
• F16 for the full precision of the model

Acknowledgements

• Very Big thanks to Jackrong [https://huggingface.co/Jackrong] for providing such good models. Without them, this model could not have been created possibly.
• Thanks to Unsloth AI for providing such high quality Quants to be used.
• This model couldn't have been created without Qwen Team for providing the base variant of the Qwen 3.5 models in such permissive licenses