😺 GPT-5 changes EVERYTHING

PLUS: Apple's making a humanoid robot?!

Welcome, humans.

This is wild: while Apple abandoned its long-rumored Apple Car project awhile back (boooo), it now looks to be developing a humanoid robot (wooo!). They already produced a framework called EMOTION to help robots, take a guess, understand human social cues and respond with appropriate gestures (paper).

Uhhh, could us humans ALSO get a copy of that? Asking for a friend…

Not only that, Apple has a bunch of other robotics ideas in development, like this research paper exploring a robot version of the Pixar Lamp. Another project in the works is a tabletop robot sorta like an iPad on wheels.

Idk about you, but we want our future Apple bots to be able to do what Carnegie Mellon and NVIDIA researchers’ new ASAP framework can do (paper):

ASAP: Aligning Simulation and Real-World Physics for Learning Agile Humanoid Whole-Body Skills

Their Unitree G1 robot nails Kobe's fadeaway jumper, recreates LeBron's “Silencer” celebration (complete with the single-leg landing!), and even pulls off Ronaldo's signature 180° celebration jump. Imagine buying a bot to play basketball with. It’d almost be like having real friends!

Fun fact: They broke two robots during training. But as they say, no pain, no gain…

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

Sam announced “GPT 4.5” in a few weeks, “GPT 5” in a few months.
Anthropic’s new model to arrive “in the coming weeks.”
EU planned $200B+ AI investment with chip factories.
Study found ChatGPT power use could be 10x lower than estimated.

Advertise in The Neuron here.

Sam just released the OpenAI roadmap for the next few months… here’s what it means for YOU.

You might not remember this, but there used to be a time when ChatGPT was just...ChatGPT. No model picker, no settings to fiddle with, just pure AI magic.

Well, OpenAI is FINALLY fixing the confusing mess of GPT models—and making GPT-5 free(ish).

In a surprise announcement, Sam revealed OpenAI's detailed roadmap for GPT-4.5 (Orion) and GPT-5:

GPT-4.5 (Orion) will arrive in a few “weeks” as the last “traditional” model.
GPT-5 will follow in a few “months”, and will include o3 on top of it.
Separately, OpenAI announced o1 and o3 mini now support files + images.

The interesting thing here is how OpenAI is simplifying by unifying everything—GPT-5 will come fully equipped with voice chat, image creation, search, and Deep Research in one seamless interface.

Here's what this means for you: No more juggling between different models or tools—just tell GPT-5 what you want to do, and it'll handle the rest.

Want a weather update? Instant answer.
Building a website? It'll methodically write and test the code.
Need research help? It'll search the web and cite sources.

GPT will technically do all of this now…but it takes some *~finesse~* to pull off.

Now, the pricing shakeup is even more interesting:

Free: Unlimited use of GPT-5 at “standard intelligence.”
Plus users: “higher intelligence” version.
Pro users: “even higher” intelligence—we talking AGI here, Sam?

But wait, there's more. Sam also had good news for Deep Research—plus and free users will get access to it:

Plus: 10 searches/month (scaling up over time).
Free: 2 searches/month.
Pay-as-you-go option planned, too.

Since Deep Research was reserved for the Pro level of ChatGPT, many of you probably haven’t used it yet—but now that it’s opening up, we’ll share some advice on how to use it in our “Prompt Tip of the Day” below.

Our take: OpenAI is finally admitting what we've all been thinking—their product lineup got waaaay too complicated.

Using ChatGPT lately feels like playing a slot machine—you keep pulling different levers (GPT-4o, GPT-4o with scheduled tasks, o1, o3-mini, o3-mini-high), hoping you'll hit the jackpot with the right combination.

Just look at this mess.

Fun fact: Three Mile Island’s near-miss disaster happened because of confusing controls—user friendly tech saves lives!

After years of increasing complexity, making GPT-5 both free and simpler is an OBVIOUS play to make AI more mainstream. We’ll see if it works!

FROM OUR PARTNERS

💼 Want to become an AI Consultant?

This stat is WILD: The AI consulting market is projected to explode from $6.9 billion to $54.7 billion by 2032. Want to capture this opportunity?

So how do you, an AI enthusiast, become a handsomely compensated AI consultant?

Well, The AI Consultancy Project from Innovating with AI has already helped 300+ students build successful consulting businesses.

Current students love:

Tools and frameworks to find and serve clients.
6-month roadmap to building a 6-figure business.
Methods to land your first client in as little as 3 days.

As a Neuron reader, you get early access to the next enrollment cycle!

Take advantage of the AI boom—request early access to The AI Consultancy Project ASAP!

Prompt Tip of the Day

Want to get the most out of Deep Research? Start by telling it you want expert-level analysis with multiple high-quality sources, define your exact research question, ask for synthesis across sources, and always demand proper citations.

Here is our complete Deep Research prompt template to try yourself.

Treats To Try.

*Wispr Flow captures your natural voice at 3x the speed of typing—no more typos, no more sore wrists. Just speak, and Flow delivers perfectly formatted text in any app. Try Wispr Flow today.
Dust lets you search and automate work across all your tools by connecting AI agents to your existing data.
Whisk from Google is now available in 100+ countries—it lets you upload a subject, scene, and a style and then remixes all three into a new image (demo).
Adobe Firefly is now available at a standalone subscription that includes image, video, and audio tools—it’s $9.99 a month for unlimited images and 20 videos, $29.99 for 70 videos, and eventually, there’ll be a 500 video price tier.
fal lets you create instant videos and product demos 10x faster than normal tools (raised $49M).

See our top 51 AI Tools for Business here!

*This is sponsored content. Advertise in The Neuron here.

Around the Horn.

Epoch AI did a study that found ChatGPT uses 0.3 watt-hours per query, which is 10 times less than previously estimated and equal to a few minutes of LED bulb usage.
- This is a big deal if true, because another report found AI data centers could need 68GW, or nearly the entire power capacity of California from 2022, in only two years time.
Anthropic plans to release a new Claude version “in the coming weeks” and projects to make ~$2-3B in revenue this year, with plans to hit $12-35B by 2027.
The EU will spend $200B+ on AI, with $20B+ specifically for five new AI “gigafactories” that will each use 100K AI chips from the last generation.
PwC said over 79% of its 75K US employees actively use GenAI tools in their daily work, with an average of 26% of people using AI at least once per day.

FROM OUR PARTNERS

The one* AI tool we use every week…

We (Noah & Grant) use ChatGPT + Claude every week, but we ALSO use a killer product called Attention.

It’s a must-have for all sales teams, so if you sell literally anything, drop what you’re doing and go book a demo with Attention ASAP.

Thursday Trivia

One is a real, and one is AI. Which is which? (vote below!)

Which is AI?

The answer is below, but place your vote to see how your guess compares to everyone else (no cheating now!)

Here are the results from last week’s trivia (B was AI):

8 humans, 4 robots

Here’s what you said:

A.L. chose A: “Windows in the background look improperly textured as well as shadows on vehicles.”
E.J. chose B: “I'll play! The skin's smoother than a fresh jar of mayo, but the eyes tell another story—like an AI couldn't decide on her age. Her forehead says 20 (and no, it’s not Botox or a filter), but her eyes have wrinkles like she’s in her mid-40s.”
L.R. chose A (but meant B): “I accidentally hit the wrong button The top image is flat and not great but realistic The ai image is super fake. Not everyone looks like a model.” So true bestie… justice for us normies!