An email came from Ollama saying that a new cloud model has been released.
They say there are many performance improvements compared to the existing m2.7
Minimax said that they are good for repetitive tasks???
Usually when coding, you do a lot of repetitive tasks, so compared to expensive things like Claude, Codex, GLM, it seems they were saying theirs is cheaper and more advantageous
It seems like while upgrading, they wanted to create a model with powerful performance like models from other companies
If the model is powerful, it might be good for design, but instead tokens just keep melting away
Honestly, I'm not sure if m2.7 has cheap tokens
It seems to be better than DeepSeek v4 flash, but.....
