> I just shelve it 'til next month and the next big OpenAI/Anthropic/Google model will usually crush it.

1 month to write some code with LLM, that's quite the opposite of the promised productivity gain