I ran OpenAI’s new GPT 5.2 through a three round test. It still scores below Grok and Gemini.
Some of GPT-5.2’s responses are excellent. But the quality of its answers are inconsistent.
Let me show you where this model excels and where it falls short…
Round #1: Learning About Needle Biopsies
Tomorrow, I’m meeting with a company that’s working on cancer diagnostics. They’re going to make needle biopsies easier.
Can GPT 5.2 help explain how needle biopsies work?
GPT 5.2 gave a fantastic answer here. It clearly explained exactly how needle biopsy and MRI work together.
It also noted that some needle biopsies are done together with CT scans, which I did not know. When I clicked through on the citation, I found a high-quality source from the National Institutes of Health.
I’m giving this round an A+!
Round #2: Investigating AI Market Research Startups
This afternoon, I’m meeting with a startup doing AI market research. It creates synthetic audiences to test marketing content.
So I asked GPT 5.2 for an overview of the startups in this area. I asked for a ton of information, including product features and the amount of venture capital they’ve raised.
GPT 5.2 crushed it here, giving a great summary and lots of detail on the leading startups. When I clicked on citations to verify how much funding these startups raised, the information checked out.
I’m going to give this round an A+ as well!
Round #3: CoachGPT
For our third and final round, I asked GPT 5.2 to help me with my workouts.
My goals for weight training are hypertrophy and longevity. What exercises should I try?
GPT 5.2 gave a nice response with useful pictures. It recommended some exercises I’ve never heard of, like the “dead bug” for core strength.
But it did not cite a single source. So, it’s hard to know how good this information is.
Are these really the best exercises for my goals? Or are they just random things GPT 5.2 came up with?
I also would like to see GPT 5.2 respond with some videos, not just still images. When it comes to learning an exercise, there’s nothing like a video.
I’m giving this round a C.
Wrap-Up
Overall, GPT 5.2 earned a B+ in my testing.
It seems to be tuned toward white-collar work, like investing and medicine. OpenAI alluded to this in their announcement.
But when I use it for anything else, GPT 5.2 is pretty weak. It wasn’t very helpful in helping me plan a workout.
Our goal is Artificial General Intelligence. AGI would give great answers to all sorts of questions.
GPT 5.2 doesn’t do that.
OpenAI’s latest model isn’t bad, but it’s still behind Grok and Gemini. Grok scored an A in my testing and Gemini scored an A-.
No wonder Sam Altman called the code red. OpenAI may have started the AI revolution, but today, it’s falling behind.
Have you tried GPT 5.2?
More on tech:
Grok 4.1 Thinking Beats Gemini 3.0 Pro in Real World Test
Can DeepSeek Beat the Best American Models?
How to Make a Great AI Product: Oboe
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.
Tremendous 
Leave a comment