I Put Elon’s New Grok 4.3 Beta to the Test — Here’s Why It’s My Favorite Model So Far

+

Elon has quietly released Grok 4.3 Beta, xAI’s most powerful model yet. I tested it this morning, and it’s the best model I’ve ever used.

I put 4.3 Beta through a grueling series of tests this morning. I had it analyze 5,000 venture deals, predict the future of the real estate market, even find under-the-radar travel recs for my trip to Tokyo.

Elon’s new model excelled across the board. Let me show you what the new Grok can do…

Round #1: Real Estate Investment Outlook

High interest rates have meant a tough couple of years for real estate. That includes my Fundrise portfolio, which invests in housing across America.

What do the next couple of years have in store? Let’s ask Grok…

Grok predicts that Fundrise will see 8-15% returns over the next couple of years. This would be a big upshift from the low single-digit returns of late.

Gradually declining interest rates should support real estate. Strong residential demand also plays into Fundrise’s strategy, which is heavy on multifamily.

Grok cited solid sources like CBRE and Reuters. Its analysis was clear and easy to understand.

Compared to the prior model, Grok 4.3 produced a more readable response. It’s shorter with better formatting, making it easier to digest.

I’m giving this round an A!

Round #2: Planning My Trip to Tokyo

A week from today, I’m getting on a plane to Tokyo! Lately, it’s all I can think about.

It will be my first visit to Japan in six years. Ever since I first visited in 2014, Tokyo has felt like my home away from home.

Can Grok recommend some awesome, under-the-radar spots for me to try?

I’ve actually been to around half these spots, including Todoroki and Hamarikyu! In each case, Japanese people brought me there.

This tells me that Grok’s recommendations are superb. They’re the type of under-the-radar spots actual Tokyo residents recommend.

The only flaw is a lack of dining options. Grok recommended some beautiful natural areas, but not much in the way of food.

I’m giving this response an A-.

Round #3: Analyzing 5,000 Venture Deals

Back to business…

For the final round, I gave Grok 4.3 a difficult task: analyzing the nearly 5,000 deals I’ve passed on in the last couple of years. Can Grok find mistakes and opportunities to improve?

I fed Grok a monster spreadsheet that includes details on every deal. Here’s what it found…

Grok thinks my approach is solid, carefully hewing to my thesis. I’ve systematically filtered out companies without builder founders, products or customers.

My biggest opportunity to improve: tracking the close calls.

Now and then, I find a company that almost meets my bar for investment. I haven’t systematically tracked the outcomes of those companies. That’s something I plan to do more of, assisted by AI.

Grok took its time, running analysis for around eight minutes. Still, that’s nothing for the amount of data I threw at it!

Grok’s analysis was thorough and helpful. I’m giving this round an A+.

Wrap-Up

Grok 4.3 Beta earns an A overall in my testing.

It’s not a quantum leap from Grok 4.20. But its analysis is tighter, its formatting better, and its outputs more usable.

That makes Grok 4.30 Beta the best AI model I’ve used yet. If you’re looking for a frontier model, Grok 4.3 Beta is it.

For now, 4.3 Beta is only available to paid Supergrok subscribers. Plans start at $30 a month. It’s not available via API yet, but I expect that to be released soon.

Everyone is excited about Anthropic right now. But xAI is building some incredible stuff in the shadows.

Don’t be surprised Elon if leapfrogs Dario soon…

I Put Elon’s New Grok 4.3 Beta to the Test — Here’s Why It’s My Favorite Model So Far

Share this:

Leave a comment Cancel reply