Generally is good
Generally is good, in some cases make stupid mistakes, which I have to correct by prompt, but each generation cost coins. If you construct prompt with other LLM, then everything fine. I'm happy, but needed time to understand in which cases the model make mistakes.