Hey Reader,

Google dropped Gemini 3 Pro this week.

Everyone lost their minds.

Then 24 hours later, OpenAI was like: "F*ck you Google! Here's our new model and it's even better!"

So here's what happened in these 24 hours of model insanity.

And more importantly: If you should even care.

What's Really Going On

I've watched this play out for a year now.

Every single time a new SOTA (state of the art) model drops, the same thing happens:

Before release: "This is going to change EVERYTHING!"

After release: "This is garbage. Total disappointment."

This, I'm afraid, is pure skill issue, and here's why.

Every new model has its own personality.

Like a new coworker who does things differently than the last guy.

It takes a minute to figure out how they work best.

But people don't want to wait. They want instant perfection.

So they trash it and move on, just to eat their words weeks later.

My Take on Gemini 3 Pro

The model is actually really good. Especially for design work.

But that's not even the big story here.

The real question: is it better than what we're already using?

And naturally, I don't know yet. Nobody does.

That's why you won't see a video from me saying "THIS IS THE BEST MODEL EVER" because I literally have no firm opinion yet.

What I can tell you right now:

The model works. It's solid. But "better"? That takes weeks to figure out.

Not just for me, but for the makers of Cursor and other tools to optimize their systems for it.

Until then, nothing really changes for you unless you want to experiment with it.

Then OpenAI Dropped The Bomb

24 hours after Gemini 3 Pro released, OpenAI introduced GPT-5.1 "Codex Max"

Ridiculous name. The memes have been incredible.

When I first read about it, I got excited. Running for 24+ hours on single tasks.

Supposedly the most autonomous model yet.

Here's where they screwed up though...

They added "significant speed improvements" which... sound good, right?

Wrong.

The model started behaving less intelligently. Making worse decisions.

That's not what you expect from Codex models.

If you wanted interactive speed, you'd use Sonnet 4.5, Composer-1, or now Gemini 3 Pro.

Codex models are supposed to be smart and autonomous.

They make quality decisions on complex problems that you probably can't.

I don't understand where Codex Max fits.

But I'll test it as soon as it's available in Cursor (which the team confirmed will be soon).

The Bottom Line

Don't listen to people with opinions about a model that just dropped.

Don't let anyone tell you to start using a brand new model because "it's killing X" or "replacing Y" – they're full of sh*t.

I will test it and let you know. That's my job. To reduce your tool overwhelm.

You keep building. I'll keep testing.

Talk next week,
Rob