Hey there! This is a model created by a 15 year old AI enthusiast by merging Jackrong/Qwen3.5-4B-Neo and Jackrong/Qwen3.5-4B-Claude-4.6-Opus-Reasoning-Distilled-v2 using Dare Ties one-shot merge.
This model is not yet benchmarked, but you may benchmark it and compare with the base models; and also merge this with other models to increase capability.
This model was trained on Kaggle using 2 x T4 GPUs which comes under it's free tier. Z.ai's GLM-5 also helped me code the configurations for the model(I'm still learning!)
This model may hallucinate sometimes as it stays an autoregressive reasoning model, but this is strong in reasoning, try it if you don't believe!
Please discuss the flaws of this model so I can attempt to fix it if I can.
Thank you for testing it out, it means a lot to me! 😄
