[journal]

I'm so glad I was able to use LLMs to finish my project in the sweet spot between 23 and 25 when the pricing was cheap and the getting was good, even though we had some bumps along the way had to wait and downtime. those who benefited the most are the ones who were able to earn experience using these things. cause going forward it'll only get more expensive. gpt 4.1 - 5.1 was $2 diff. we're now .1 generation @ $30 per mil. double. they've already done their projections, people need to understand this is going north of $10000/year for general purpose access.