Google recently released the new Gemini 2.5 Pro and 2.5 Flash models. So how good are the latest and greatest from Google? This morning, I ran them through 3 tests to find out.

Round 1: Find Me a Deal on Shoes

If you read Monday’s blog, you know I’ve been looking for a new pair of New Balances. And unlike a normal person, who just finds them on Amazon and buys them, I like to comparison shop.

Plain vanilla Google gave me some pretty good results on Monday, while ChatGPT flopped. Let’s see what Gemini can do!

For this query, I used 2.5 Flash, since it shouldn’t require extended thinking. At first, I just put in “men’s New Balance sneakers size 11.” It gave me a product description, which is useless.

So, I refined the query and said “find me the best price on a pair of men’s size 11 New Balance sneakers.” Let’s see if it finally works…

It still just gives me a bunch of useless AI slop. “Shop around,” Gemini tells me. No kiddin, buddy!

I’m giving this round an F.

Round 2: Researching a Startup

For the next round, I turned Gemini lose on a startup I’m researching. They sell a SaaS product to midsize financial institutions.

I used 2.5 Pro Experimental for this one. Here’s my prompt: “How many banks and credit unions between one and ten billion in assets exist in America?”

Gemini gave a good answer, including specific numbers and breaking them down by type of institution. But it didn’t cite any sources, so I have no idea if the numbers are accurate.

Put that same query into Grok, and it cites 25 sources. You can click through and verify everything Grok says.

Gemini’s answer would’ve impressed me a year ago. But today, it’s table stakes.

I’m giving this round a B-.

Round 3: Stats on Startup Success

Okay, I’ll give Google one more chance to redeem itself. I want to know about the rates of success for startups.

I used 2.5 Pro Experimental for this query as well, because it may require more in depth search and thinking. Here’s the prompt: “How many startups make it from raising a preseed round to $10 million ARR? How about to 25 million ARR and 100 million ARR?”

Let’s see what Gemini comes up with…

Hey hey, Gemini is looking alive here! This was a pretty solid response, showing that 13% of startups make it to $10M ARR within 10 years. It cites a high quality source for it — I went and verified the number from the ChartMogul report that Gemini cited.

It isn’t able to give us any stats on startups getting to $25M or $100M, unfortunately.

I put the same query into Grok, and Gemini’s response was actually better. Grok gave estimates with little basis in data that were extremely high and probably wrong (20-25% of pre-seed startups hitting $10M ARR).

Gemini gets an A on this one!

Wrap-Up

Oddly enough, Gemini did best on a harder question in Round 3 and worst on a simpler question in Round 1! Go figure.

Sometimes, what’s easy for us humans is hard for AI, and vice versa.

Averaging these grades, Gemini gets a gentleman’s C.

This is not impressive for a multi-trillion dollar company that invented generative AI. If Google wants to stay on top, it has to do better than this.

That said, Google has also released some great products. Deep Research is really impressive and almost on par with Grok 3, which is the best I’ve used.

With a lot of smart people and unique data, I wouldn’t count them out.

More on tech:

Using Grok 3 to Manage My Stock Portfolio

ChatGPT’s New Shopping Tools: Better than Google?

DeepSeek vs. Gemini Deep Research: Which Model Is King?

Save Money on Stuff I Use:

Fundrise

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!

Misfits Market

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.