Welcome, humans.

Happy Friday! These candid selfies of famous video game characters are totally awesome:

There’s a ton of other great ones in the comments, though gotta be real with you—this Pacman one is straight nightmare fuel.

Also, IDK whose old enough (or young enough) to remember Katamari Damacy (the game where you roll around and pile entire cities into a giant ball), but this selfie is such a throwback, we’re obsessed.

Another weird trend like this: asking ChatGPT what your car would look like as a person. The Cybertruck straight up SENT us.

A NOT so fun trend? o3 and o4-mini’s incredible ability to reverse location search photos. Don’t get us wrong, that’s an incredibly useful feature…but with great power, comes great irresponsibility…

Here’s what you need to know about AI today:

• We compare how o3 and o4’s “genius” stacks up against Gemini 2.5.

• Google offered free AI Premium to college students until 2026.

• OpenAI almost purchased AI coding startup Cursor before a deal with Windsurf.

• Scientists mapped the largest 3D mouse brain reconstruction

Let’s talk about that “o3 = genius level AI” claim (and compare o3 to Gemini 2.5)

So OpenAI has two new AI models out—should you use them, or nah? The answer really depends on whether you want the best of the best, the best for the price, or just good enough.

TL;DR: Best of the best = o4 mini, Best for the price = Gemini 2.5, just good enough = whatever brand of model you prefer at the $20 a month tier, cause they’re all competitive.

Now, you might be wondering: are these new models actually “genius?” Hmm, wherever did you get THAT idea?

Oh, right. This X post from Sam A, which has become a bit of a lightning rod for those debating o3’s over or under-hyped-ness:

Obviously, Sam is just quoting immunologist Derya Unutmaz here—but simply reposting something doesn’t make it so.

So IS o3 actually genius level? Pretty much everyone agrees the new OpenAI models are solid—maybe a tad underwhelming for some, given the hype—but genius?

Cause if they’re genius, that means you DEFINITELY should use ‘em, right?

Let’s find out.

Here’s the evidence FOR genius:

First, Derya shared o3’s results on the Mensa Norway IQ test, and it’s neck and neck with Gemini 2.5, hovering around “genius” level (136)—which to him, proves his point.

We also have the official data from Artificial Analysis on o4-mini (we know, that’s not o3 [coming soon], but it’s close enough):

• o4-mini scored a 70 on overall intelligence, beating out Gemini 2.5 Pro Preview (68) and Grok 3 mini Reasoning (67).

• It made significant gains in coding intelligence, achieving #1 in their coding index.

• It also showed impressive performance on math benchmarks, matching or exceeding others (thanks to that tool use no doubt).

Now, the evidence AGAINST genius:

AI Explained (one of our favorite Youtubers) looked at o3 and o4 mini versus Gemini 2.5 in a great head to head comparison:

As far as genius is concerned… here’s what he found:

1. While o4-mini excels at academic benchmarks (especially math and coding), it still lacks the common sense that we'd expect from a true genius.

2. He pointed to simple reasoning failures in his simple bench tests, like a test where o3 couldn't correctly count line intersections in a diagram.

3. In another test, o3 couldn't understand that a glove falling from a car on a bridge would land on the bridge, NOT in the river below.

The bottom line? These new OpenAI models represent significant progress but calling them “genius” is a stretch. They still lack some of the common sense reasoning and creative leaps that define actual, human, genius.

Then there’s the price: Gemini 2.5 Pro is significantly cheaper than o3. As AI Explained explained, the new models perform slightly better, but cost ~4× more than Gemini 2.5. As Lisan al Gaib said on X, “if you’re rich, go for o3. If you’re not, go for Gemini 2.5 Pro.”

Now, how do o3 and Gemini 2.5 stack up overall? We put GPT o3 and Gemini 2.5 head to head across 9 prompts in this full comparison. Check it out!

FROM OUR PARTNERS

Your AI product is smart. Your pricing should be, too.

You’ve built something powerful with AI. Now it’s time to get your monetization right. Join pricing experts from 49 Palms Ventures and Metronome CEO Scott Woody for a live discussion on how to design pricing that reflects your product’s value and drives growth.

You’ll learn how to:

• Design your pricing with a 9-step framework. 

• Clearly map pricing to product value.

• Navigate your first sales deals.

Register for the webinar and save your seat here

Prompt Tip of the Day

Did you know you can combine multiple images into one image with ChatGPT’s image generator? Here’s a good example:

Treats To Try.

1. tl;dv takes your meeting notes while you run the show, automatically updating your CRM and drafting follow-ups—free forever.

2. Otter (AI notetaker) can apparently add live captions to your Zoom meetings by connecting to Zoom's captioning system—free to try.

3. Veo 2 (Google’s leading video generator) is now officially in the Gemini App.

4. The Librarian manages your emails, calendar, and documents across all your platforms (G Suite, WhatsApp, Slack)—free trial, then paid.

5. Wiza Monitor alerts you when prospects change jobs so you can reach them first with their new contact details—free trial, then $99/month.

6. Polarr Next lets you edit thousands of photos instantly by learning your style from a few reference edits, cutting your workflow by 80% while keeping your files offline—free to try, then $19.99/month (yearly); check out their Color Match tool.

7. Omakase Voice turns your website into a voice-powered sales agent that listens, talks, and recommends products in real-time—free to try.

8. Vapi adds voice AI capabilities to your automation stack, allowing you to handle inbound and outbound phone calls that integrate w/ workflow tools like Make and n8n (to automate anything)—free trial available.

9. Trueguard stops fake users from abusing your SaaS by instantly detecting suspicious emails, VPNs, and bots—free to try.

10. Hence Global tracks global risks in real-time and tells you exactly what to do about them, saving you from expensive consulting contracts—paid only rn ($1.5k/ year), read more here.

See our top 51 AI Tools for Business here!

Around the Horn.

• Google suddenly woke up to OpenAI’s genius idea of offering free AI to college students, and will offer their One AI Premium plan (typically $20) for free until June 30, 2026—talk about getting ‘em hooked.

• The LM Arena (the O.G. AI leaderboard) will form a company to continue improving its service, which crowd sources user feedback on AI responses to rank them.

• OpenAI launched a new Flex processing option that cuts your AI model costs in half by handling lower-priority tasks with slower response times (reducing o3 costs from $10/M to $5/M input tokens, for example).

Intelligent Insights

• OpenAI apparently almost bought Cursor before deciding to hold talks with Windsurf.

• Timothy B. Lee of Understanding AI wrote this great piece on where frontier AI models are today, and why it’s getting more difficult to test them.

• Road to Artificia argues copyright won't stop AI progress because governments will prioritize technologies making intelligence widely available over copyright systems that have abandoned their anti-monopoly origins.

• Check out this field guide to AI product improvement (based on 30+ implementations) that shows how the most successful AI teams obsess over measurement and iteration rather than special tools and frameworks.

• Researchers have created the largest ever 3D reconstruction of a mouse brain (paper) mapping both structure and neural activity, revealing brain algorithms that could transform AI by showing how evolution solved efficient learning with minimal energy—potentially helping develop systems that match the brain's ability to quickly learn and generalize from limited examples.

A Cat's Commentary.