If you read the blog post, you will actually see the weights mentioned.

There is no point arguing with the "it just predicts next word" crowd. They only look at what an LLM technically does, and while they are of course technically correct, they completely ignore emergent capabilities, speed of progress and societal impact.

The next discussion to have is not whether we have achieved early stages of AGI, but whether it matters. As long as we're not pretending that a system is sentient (which is a separate issue from whether it has AGI properties) it ultimately doesn't matter how it reliably solves a multitude of problems as if it had general intelligence; it only matters that it does.

If you go to all this trouble, please keep in mind that electricity prices vary a lot across the world. In some places in Europe people pay twice as much as you assumed above.

Making it clear how you arrive at your value calculation in an updated post (or even making it a dynamic calculator where people can enter their cost/kWh) would be very useful.