My first impressions of Minimax M2.7 (Q5_K_M) vs Qwen 3.5 27b (Q8_0)
I'm not sure if the AesSedai's Q5_K_M version of Minimax M2.7 is too much lobotomized or if the model itself is kind of weak. I did a simple experiment with both models running with the recommended parameters.