Anthropic just dropped Claude Sonnet 5. Bigger context window. Better at coding. Closing the gap on their top-tier model. All the headlines are saying the same thing: this changes everything.

Here’s what nobody in the AI hype machine wants you to notice: every single time a new model drops, they tell you it changes everything. Every time. For two years straight.

And sure, technically, it’s better. The numbers are real. But better at what, for who?

If you’re building rocket ships or training your own AI, sure, you probably care about squeezing out that extra 5% performance. But if you’re like most people using AI to write emails, plan content, answer customer questions, or organize your business, you hit “good enough” a long time ago. You crossed that line probably a year ago and didn’t notice.

The AI companies know this. That’s exactly why the game has quietly changed. Notice what’s actually happening underneath the “most powerful ever” headline: the price dropped too. A lot. Because they’re not really competing on who’s smartest anymore, they’re competing on who’s cheapest for something that’s already smart enough.

That’s the part nobody says out loud: the AI arms race stopped being about intelligence and started being about price.

Which means the real move for you isn’t “upgrade to the latest, biggest, most expensive thing the second it drops.” The real move is asking one boring question: is what I’m using right now actually failing me at anything? If the honest answer is no, you don’t need the new one. You need to stop scrolling AI news like it’s a shopping list.

I say this as someone who genuinely loves this stuff and tests new tools constantly. But there’s a difference between staying informed and getting played by a marketing cycle that’s designed to make you feel like you’re falling behind if you’re not on the newest thing.

Here’s the only real reason to actually switch: if the new version is meaningfully cheaper for the exact same result you’re already getting. That’s it. That’s the whole test.

Everything else is noise dressed up as progress.

So yeah, Sonnet 5 is impressive. Good for them. But before you go rebuild your whole workflow around it, ask yourself what it’s actually going to do for you that your current setup isn’t already doing. If you can’t answer that in one sentence, you already have your answer.