Zuck is investing $70 billion in AI this year. Is it finally paying off? My testing says no.
When I tested Meta’s Llama model in May it scored poorly, earning a C+.
Since then, Zuckerberg created the Superintelligence team, headed by Alexandr Wang. Meta has poached top researchers with pay packages as high as $1.5 billion.
This morning, I gave Llama another shot, running it through 3 real world tests. It scored even worse, notching a D+ overall.
Let me show you how flawed Llama really is…
Round #1: Missile Defense Startups
Last week, I watched House of Dynamite on Netflix. It shows what could happen if an adversary launched a nuclear missile at the United States.
That got me thinking about missile defense. So, I asked Llama to find me the best new startups in this area.

Llama gave me a nice list of missile defense startups. But it didn’t cite any sources.
If I want to verify information about any of these companies, I need to go to Google or Grok. So, there’s no point in using Llama. It also gave me mostly late stage startups, despite my instructions to focus on early stage.
I’m giving this round a C-.
Round #2: Robots Building Robots
I saw a tweet yesterday about Chinese robots that build more robots. In a matter of weeks, the number of robots could double.
So I wondered, what startups are working on this?

Llama gave me a list of general robotics companies like Figure and Boston Dynamics. These companies make some awesome products, but they’re not focused on robots-building-robots.
Llama also failed to cite any sources. How can I rely on this information?
I’m giving this round a D.
Round #3: Llama as Personal Trainer
I work out 5 times a week, doing a mix of weights, yoga, biking and swimming. But I’ve never really been a runner.
The NYC Marathon on Sunday got me thinking: should I add running to my routine?

Llama noted that although we worry about running hurting our knees, moderate running can actually strengthen the muscles and stabilize the joint.
Llama cited a source for that info, but when I clicked through, the source had nothing to do with knee arthritis. Instead, it covered how to train for a 10k.
Once again, it’s hard to trust Llama’s info when it can’t cite a source properly. I’m giving this round a C.
Wrap-Up
Llama scored poorly in this re-test. I’m giving it a D+ overall.
That’s down a full letter grade from my test in May. At this point, Zuckerberg doesn’t have any more excuses.
He’s investing tens of billions of dollars into AI. He’s hiring the best people in the world.
Why can’t they produce a decent product?
I’m hoping Meta gets its act together and makes a model I actually want to use. Until then you’ll find me on Grok.
More on tech:
Wispr Flow: Voice Dictation That Finally Works
Can DeepSeek Beat the Best American Models?
Neo Launches Android — Which Jobs Are Safe, Which Are Toast?
Save Money on Stuff I Use:
This platform lets me diversify my real estate investments so I’m not too exposed to any one market. I’ve invested since 2018 with great returns.
More on Fundrise in this post.
If you decide to invest in Fundrise, you can use this link to get $100 in free bonus shares!
I’ve used Misfits for years, and it never disappoints! Every fruit and vegetable is organic, super fresh, and packed with flavor!
I wrote a detailed review of Misfits here.
Use this link to sign up and you’ll save $15 on your first order.
Leave a reply to OpenAI Behind Competitors Despite GPT 5.1 Release – Tremendous Cancel reply