Modern Creator
Alex Finn · YouTube

Google Gemini 4 Neon Is the Greatest AI Model Ever. Here's How to Use It

A 20-minute argument that Google's Gemini 4 family wins on writing and original thinking, which tier to buy, and why the Gemini app beats Antigravity for now.

Posted
2 days ago
Duration
Format
Review
hype
Views
36.2K
812 likes
Big Idea

The argument in one line.

Gemini 4's real edge is not coding benchmarks but human-quality writing and forward-looking, non-average answers, which makes a long focused conversation in the Gemini app the best way to use it.

Who This Is For

Read if. Skip if.

READ IF YOU ARE…
  • You pay for Claude or ChatGPT and want to know whether a Gemini subscription adds anything different for planning, strategy and writing.
  • You are confused by Google's model names (Carbon, Neon, Hydrogen), plan tiers and apps, and want a plain recommendation on what to buy.
  • You use AI mostly for writing, business planning and personal decisions rather than building 3D demos or games.
SKIP IF…
  • You want benchmark tables or head-to-head coding tests; the video deliberately skips them and shows one demo.
  • You need a verified, neutral comparison; this is one heavy user's strong opinion and he has contacts at Google.
TL;DR

The full version, fast.

The host claims Google's Gemini 4 models are the best available: Carbon is the smartest tier, Neon the efficient middle, Hydrogen the fast one. His case rests on human-sounding writing, a different way of thinking from Claude and ChatGPT, and fewer refusals on personal questions, with coding treated as a given. He recommends the $99 Google AI Ultra plan for Carbon access plus YouTube Premium and 20 TB of storage, using the Gemini app instead of Antigravity, and running a focused hour-long brain dump. He closes by speculating that Google's fast release pace may signal recursive self-improvement.

Free for members

Chat with this breakdown — free.

Sign in and you get 23 free chat messages on us — ask for the hook, quote a framework, find the exact transcript moment, generate a markdown action plan. Bring your own key when you want unlimited.

Create a free account →
Chapters

Where the time goes.

00:00 – 00:48

01 · Intro

Bold claim that Gemini 4 beats Claude and ChatGPT, plus the promise to untangle names, pricing and usage.

00:48 – 01:19

02 · The Gemini 4 models

Carbon is the Fable-class flagship, Neon the efficient middle, Hydrogen the fast Sonnet-class tier.

01:19 – 10:29

03 · Why they are the best

Strong coding, high efficiency, human-quality writing, a different way of thinking, forward-looking answers and fewer refusals.

10:29 – 10:59

04 · Slop simulation

One quick demo: a 3D tank game with destructible cities, shown to prove the coding is fine.

10:59 – 13:23

05 · Pricing

Recommends the $99 Ultra plan for Carbon, 20 TB storage and YouTube Premium; Neon works on the $20 plan.

13:23 – 15:55

06 · How to use Gemini 4

Antigravity is bare-bones with no mobile app; use the Gemini app for knowledge work and the Antigravity CLI for code.

15:55 – 17:49

07 · How the host uses it

Spends most of his time in the Gemini app, brain-dumping everything for a focused hour on Carbon.

17:49 – 20:16

08 · Something scary is happening

Speculates Google's rapid release cadence could mean it is first to recursive self-improvement; asks viewers for feedback.

Atomic Insights

Lines worth screenshotting.

  • Most chat models give you the average of the best past answers; the host's claim is that Gemini 4 feels like it predicts what will work in the future instead.
  • Claude and ChatGPT tend to answer the same question the same way, so adding a model trained differently gives you a perspective the other two can't.
  • The host says Gemini refuses and deflects to professionals less often on personal questions than Claude or ChatGPT, which he treats as a feature for unusual challenges.
  • Coding is no longer a differentiator: every top model is good at it, so the interesting comparison is writing quality and how a model reasons.
  • Demos of 3D simulations and pelicans on bikes are popular because they are visual, not because they reflect what most people use AI for.
  • Only the roughly $99 to $100 Google AI Ultra tier includes Carbon, while Neon is available on the $20 plan.
  • The Ultra plan bundles 20 TB of storage and YouTube Premium, so the host argues it is a better value than a $200 AI-only plan.
  • Google has no standalone Gemini CLI anymore; the coding command line is the Antigravity CLI.
  • Antigravity is described as a bare-bones coding harness with no mobile app, so the host uses the Gemini app for knowledge work instead.
  • A suggested exercise: close every other app and talk to the model about your business, career and personal challenges for one uninterrupted hour.
  • The host's recursive self-improvement guess rests only on release cadence: four Gemini releases in about four weeks versus roughly two months per release elsewhere.
  • Google's confusing lineup of apps and plans is blamed on a huge, splintered organization with hundreds of teams.
Takeaway

Pick a model for how it thinks

WHAT TO LEARN

When top models all code well, the useful question becomes which one writes like a person and reasons differently from the others you already pay for.

  • Coding quality has converged across top models, so choose by writing style, reasoning, refusal behavior and cost per unit of useful work.
  • Two models trained differently give you different answers; using one from each lineage is a cheap way to get a real second opinion.
  • Model names are tiers: map each vendor's flagship, efficient middle and fast model to the ones you already know before comparing prices.
  • Plan value depends on the bundle: a $99 plan with top-tier access, 20 TB of storage and ad-free YouTube can beat a $200 AI-only plan.
  • Match the app to the job: use a chat app for knowledge work and brain dumps, and a CLI or harness for building software.
  • A single uninterrupted hour of honest conversation about your business and life is a better test of a model than a benchmark demo.
  • Treat claims about the future, like imminent recursive self-improvement, as speculation built on release cadence rather than evidence.
Glossary

Terms worth knowing.

Carbon
The largest and smartest model in the Gemini 4 family, compared by the host to Claude Fable. It requires the top Google AI Ultra plan.
Neon
The middle Gemini 4 model, balancing intelligence and efficiency. The host compares it to Claude Opus and says it is available on the $20 plan.
Hydrogen
The fastest, cheapest Gemini 4 model, positioned like a Sonnet-class model and the successor to the Flash tier.
Antigravity
Google's coding harness and CLI for Gemini models, described in the video as comparable to Cursor a year ago.
Harness
The app or tool wrapped around a model that gives it projects, tools, a browser and chat management.
Recursive self-improvement
A loop where an AI improves itself, ships the result, then improves again, each cycle faster than the last, leading toward a takeoff.
Resources

Things they pointed at.

10:59productGoogle AI Ultra plan
13:50toolAntigravity
15:00toolAntigravity CLI
Quotables

Lines you could clip.

05:20
“This might be that moment where we go from slop to actual quality because the writing on this is human-like.”
single bold claim, no setup needed→ TikTok hook↗ Tweet quote
16:50
“Close out Chrome, close out everything. Open up the Gemini app and have a focused conversation for an hour.”
concrete challenge viewers can act on today→ IG reel cold open↗ Tweet quote
The Script

Word for word.

Read-along

Don't just watch it. Burn it in.

See every word as it's spoken — crank it to 2× and still catch all of it. The same dual-channel trick behind Amazon's Kindle + Audible.

Yes, it's true. The Google Gemini 4 class of models are the best in the game, hands down. Better than Claude Fabel, better than Chad GPT -Astra, better than them all.
If you are not using these models the right way, you are missing out big time. Here's the issue. They're very confusing.
The names are confusing. The pricing tiers are confusing. Which ones you should be using are confusing.
In this video, I'm going to clear it all up. I'm going to tell you which models you should be using, why they are the absolute best, which price tier you should be on and why I believe something very scary is happening at Google right now.
Let's lock in and get into it. So Gemini 4 Neon is their newest model, but we are going to be covering all their models, say in the Gemini 4 class. That's neon, that's carbon, that's hydrogen.
For those who don't know, by the way, here's kind of what they equal out to. Carbon is like their fable class, their big one, their smartest one. Neon's that kind of sweet, juicy center where it's really intelligent, but also Efficient and hydrogen is that super fast one previously like known as like Gemini flash But now it's like kind of their sonnet level model where it's still pretty smart, but it's extremely extremely fast and efficient I'm gonna be going over which ones you should be using and kind of what job each our best because it's not perfectly equivalent to these three, but I'll go over that in a second.
Now, why are these models so good? Listen, I can do what all the other AI channels are doing and spend 15 minutes going through a bunch of benchmarks that make absolutely no sense that are completely faked by every company and show you 15 minutes of like slop blender 3D simulations. I'm not going to do that.
I will do that. I'll show you one slop blender simulation. We'll do that in a second, but here is why it's the best.
Let's get straight to the point here. Number one, it's fantastic at coding. It's just as good as all the cloud models at coding.
It is up there at best in class. But here's where it starts getting better than everyone else. It is super efficient.
So you are getting a ton of usage. You know, Fable was probably the goaded model before this. The issue is, is you don't get a ton of usage on that.
They like literally have a separate meter. for Fable. You can use carbon in your plans.
You use the AI Ultra plan, you can use carbon as much as you want in it. So it's super efficient. On top of that, Neon is like unbelievably efficient.
You're getting way better cost to productivity ratio on Neon than pretty much any other model out there. It is best at class in writing. And this is where I think it stands out the most and where I think you need to kind of change the way you think about AI and what you use it for.
I think if you... Again, watch any other AI channel, even mine up to this point. Every kind of demo is, oh, look at this cool 3D simulation.
Look at this car, this pelican riding a bike, this video game, right? And that's cool. That's fine.
It makes sense for videos because it's super visual, right? And it's like fun to watch video games being made. I'm a huge gamer.
The moment I'm done filming this video, I'm going to go play Grand Theft Auto 6. The issue is, is that the strengths of this model, neon and carbon, are far outside. of just 3D simulations, right?
It's actually really strong at something that's, I think, way more relevant to most people, which is writing in the way it talks, in the way it thinks. It is the best in class at writing. It sounds and feels like a human when you read its output.
This might be the one that kind of changes how we use the internet because I think up to this point, when you see AI comments, when you see AI writing, you instantly know it's AI. You instantly, when you read like an article generated by AI, you know, like it uses all the same kind of nomenclature and things like that.
This is different. This might be that moment where we go from slop to actual quality because the writing on this is human -like. The writing on this is human quality.
It is excellent at writing. All of your writing tasks you do, if you do anything writing related, you need to instantly sign up for this plan, which again, I'll go through pricing in a second. little confusing.
I'll tell you which tiers you need to go with. But if you do anything related to writing, which I think is significantly more people than do like 3D blender simulations, then you need to sign up for this immediately. But here is where I use it the most.
Here is where... I think it's revolutionary, uh, and why I think it is a must use in, in probably a way you're not using AI as much that you should be using it is that it thinks very different. What does that mean?
What does that mean? It thinks different. I'll show you a couple of examples in a second too.
I just want to kind of go over high level, the important points here. I think when you use Claude, when you use chat GPT, they're excellent, they're excellent, excellent models. This is not me crapping on them or talking down there.
They are incredible, but to me it kind of feels in a way that Claude and ChatGPT are like trained on the same data and trained the same ways.
Like there's differences between the two models, right? Claude's a little bit better at coding and ChatGPT is a little bit better at kind of knowledge work. But at the end of the day, when you ask it questions, both of them kind of give you the same answers.
They kind of think the same way. They kind of get scared at the same types of questions, kind of deny you and push back at the same types of like requests. Gemini is completely different.
You can ask a question to Claude, to ChadGBT, and to Gemini. Claude and ChatGPT will probably give you the same answer. Gemini will give you a completely different answer than those two.
It feels like Gemini was trained on like just a completely separate set of data than Claude and ChatGPT. You know, Anthropic started when, you know, a few people from OpenAI left and then started Anthropic. And they probably took a lot of the ways that OpenAI builds models and brought it to Anthropic.
That's why there's a lot of similarities too. Gemini though, that's... demis, right?
That is people that never worked on Claude or opening eye building their own models completely from scratch, no distillation, nothing like that. And you can tell when using this model that it is just a completely different mind of its own than any other model you'll use. And again, I know this is very high level and hand wavy.
I'll go into specifics in a second, but I think this is probably the biggest reason you need to sign up for Gemini is that you are going to get perspectives that you can't get from. other AI. And for me personally, someone who plans out a lot of things, is planning my business out, is planning how to do marketing, is planning how to build products and come up with new features.
It gave me perspectives that have directly led to significantly better products and my business doing significantly better. I'll just give you an example. I am building this harness.
I'm going to open source it soon. It is a local AI model harness. It is a AI agent harness built completely for local models.
No cloud models, nothing like that. It's all about local models, helping you get up and running with local models. Gemini designed this completely itself.
It designed all the recipes, the way it works, the way it thinks, the system prompt, all of that in a way that OpenAI and Claude, they came up with a way to design this, but it was very kind of AI -like. It was very slop -like. Not that I would call Claude and Chad GBT slop, but what I mean by that is most models, when you talk to them, are very...
backwards looking. What does that mean? Backwards looking models.
Up to this point in time, when you talk to AI models, they're kind of giving you the average of all of the answers. So when I say, how do I build an AI agent harness? It kind of looks at all the other AI agent harnesses out there and then gives you the average of the best answers, right?
That's how most AI models work. And that's how I kind of feel like I get answers from Claude and Chad GBT. And listen, that's...
Awesome. That's not a bad thing. I'm not putting them down for doing that.
Because before that, I wasn't getting any sort of average of any answers. I was just pulling things out of my head, which most times not nearly as good as getting the average answer on the internet. But when using Google Gemini, neon carbon hydrogen.
It feels like the first time I'm kind of using an LLM that's forward facing. Instead of kind of looking in the past, getting all the answers and giving the average, it kind of looks forward and thinks independently based on the data. It kind of predicts based on what is going to be the best in the future on what you should do.
And I know that's kind of obscure and maybe hard to understand for people, but there's nuance there. And I kind of want you to think about that for a second. Most LLMs at this point giving you the average of the past.
It feels like when using Gemini, it's giving you like a prediction based on the future, which is very, very different than what other models are doing. It's also, I found, way less likely. than Claude and Chad GPT to push back on you when you ask about things.
So just to be honest about something, I was going through something in my personal life and I asked Claude and Chad GPT about, I just wanted to get different perspectives on this challenge I was facing. Both Claude and Chad GPT kind of skirted off. They didn't want to give me personal answers.
They didn't want to influence my life. So like, oh yeah, you should definitely, you know, seek a professional. Here's how you can find a professional.
Gemini does not, here, let me help you out. Here's kind of my perspective on things. And it didn't have any sort of boundaries or walls.
Like it's totally willing to answer all questions and give you a perspective and like kind of give confidence in its answer. Claude and Chad GBT, the moment you do something that I feel like Anthropic or OpenAI could get sued over or get in trouble for, they just, oh, no, no, no, no, no, no. Here, find a professional, find a professor, find someone else to answer the question.
Google's like, I know, screw it, I'll answer anything. What do you got? Let me help you out on anything.
So that's another reason, like if you're someone who has like very unique challenges that aren't, hey, build me a Pelican on a bike simulation. I think this is a great model to be using. Now, listen, I talked a crap ton about things that aren't vibe coding and building 3D slop simulation apps.
I'm not going to spend 10 minutes going through benchmarks that mean literally nothing. It's great at coding, right? We've gotten past the point.
All these models are great at coding. It's fantastic at coding. I am now personally, when it comes to AI, way more concerned about like the way it thinks and the way it's willing to answer questions and the way it's willing to step outside the box and be forward thinking and predict the future.
I honestly think that's way more important than 3D slop simulations. But if you're one of those people, hey, give us a 3D slop simulation. Here you go.
Let me give you a 3D slop simulation. Here is a game I built with Neon that is a 3D simulator where you can just destroy cities in your tank. All the buildings are...
Destructible you can change all your different weapons. Everything gets destroyed You can see a lot of physics and smoke effects when my tank runs over cars. I can blow things up They all have realistic destruction.
There you go. There's your 3d slop simulation I'm not gonna expose you to 20 more minutes of zombie multiplayer games. You got it.
The coating is great So let's do this Let's go through pricing and then let's go through how to use it because I think that's probably the two things people are most confused with Right now is like how do I use it and how much does this cost? I don't really get it. It is Google's biggest issue at the moment.
Like they are a little splintered. I'd say, you know, out of all these AI companies, Google has by far the most employees. And I think they're split into hundreds of different teams and those hundreds of different teams, there's probably some issues around communicating with each other.
And so that's why there's 10 billion different apps and 10 billion different plans. So let's do this. Let's go through pricing.
Then I'll go through how to use it and which of the hundred Google apps you should be on to use these models. So here, are the ones you need to know about.
I personally am on the biggest plan. There's like a $200 Google AI Ultra plan. I am on that.
I highly recommend out of all these plans, you go with the $100 Google AI Ultra plan. I think this is the best value by far out of all of these. The reason being is you get all these extra benefits on top of being able to use carbon, right?
So this is the only plan where you can use Gemini for carbon. You can use Neon on the $20 plan. Neon's great.
But if you want carbon, which I think is the best model on planet Earth right now, you need to be on the $99 plan. But on top of it, you get a bunch of things, right? So you get 20 terabytes of storage.
which is just outrageous. That is all the storage you need on the internet. So you get tons of cloud storage.
And then the next best thing you get is you get YouTube premium, which is the YouTube plan where you don't see any more ads. And I will tell you this, if you're not on YouTube premium, like you're really screwing up. It is by far the greatest subscription plan you can ever get on planet earth.
Using YouTube without ads will absolutely change your life. And I promise you, if you use YouTube for like a month without ads and then like go, on a YouTube not logged in and see what it's like with ads, you will think it's like pure torture using YouTube to be quite honest.
So you want YouTube premium. It is fantastic. So I highly recommend the $99 a month plan so that you can get all those things.
You can get carbon, you can get YouTube premium, you can get 20 terabytes of storage. That is absolutely absurd amount of storage. This is the best value there is.
I understand every AI plan I'm playing or it's now like $200 a month. This is $99. It's not only gonna take care of your AI, it's gonna take care of YouTube, it's gonna take care of online storage.
So I think this is the best value in AI at the moment. Highly encourage this one. So how do you use...
Google Gemini for carbon and neon and hydrogen. Well, how do you communicate with it? Well, basically the only harness built by Google at the moment for using these models is anti -gravity, which is what you see here.
Anti -gravity is basically what like cursor was like a year ago. And I don't mean this as any disrespect to the Google team. I am very close with the Google team.
I know a ton of their project managers. They invite me to their events. I tweeted about Google Gemini yesterday and actually one of their project managers.
called me up and we had a conversation about anti -gravity. This is going to get much better. But at the moment, this is a very bare bones coding harness.
You get your projects, you can create different chats in the projects. It has computer use. It has its own browser so you can test games and apps straight inside of it.
But there's nothing really on top of that. It's very bare bones. You don't have any sort of like dots or duet type agent in there.
It doesn't have a great interface for organizing your other chats. It doesn't have like, you know, cool things like spaces like there's in chat GPT. And the biggest downside of it all is there is no mobile app for anti -gravity.
So here's what I would say on the best ways to use these new Google models. are doing knowledge work, I would use the Gemini app. So if you're on mobile, just put the Gemini app on mobile.
If you're on desktop, you can download like a Gemini desktop app. If you're doing knowledge work, if you need tasks on your computer, that's the best way to do it. If you want to do coding, I would pass on anti -gravity right now.
Although I've talked to the, again, the team. There's many things coming, including in the next couple of weeks. So it's going to improve.
But if you're watching it at the time of this video, if you're going to code, I'd highly recommend just using the CLI. So use like the anti -gravity CLI. There's no Gemini CLI.
They canceled that. It's anti -gravity CLI. And if you want to take it up a notch from there, you can use it in Herder.
That's how I use it. Highly recommend using it in Herder. But just know moving forward, you're probably going to hit the point eventually where anti -gravity is the way to use Gemini based on how much they're improving it.
So I would download the Gemini app and just prepare for that future where this gets much better because it's crazy. For a couple of years now, Google Gemini was way behind Chad, GBT and Claude, way behind, right? And we thought they were kind of cooked Google.
We thought they were kind of dead, but all it took was them to release a couple of models that are best in class. And now they are all the way back. So I don't see any reason why anti -gravity can't get there as well.
But here's how I would use Gemini right now. I'm going to be honest. with you, I don't even code with it a ton.
I spend 99 % of my time in Gemini, in the Gemini app, just having conversation and going back and forth. I think this is the absolute stone cold best way to be using these models right now. Just open the Gemini app, brain dump everything you're thinking about right now, everything you're working on, all your lines of business, your career, your personal life, your challenges, your personal challenges, whatever it is.
Start chats and Gemini with it. Select carbon if you're on the $100 plan. Select neon if you're on the cheaper plan.
I like to use carbon. Everyone's raving about neon right now. It's super efficient, but for me, I like to have the most intelligence.
I'm willing to sacrifice some speed for more intelligence, so I actually use carbon more than neon, but it's up to you. want intelligence, you want efficiency. Neon kind of gives you both, but carbon gives you maximum intelligence.
Open up Gemini, choose the model that's appropriate for you. And I know this is tough. I know for the ADHD brain rot people out there like me, it's like super hard opening up a chat box and sitting there and chatting and focusing for an hour.
But I highly, highly challenge you. Close out all the other apps, close out Chrome, close out everything. Open up the Gemini app.
put in carbon and just have a focused conversation for an hour about all your current challenges, whether it's business side, career side, personal side, have a conversation, go back and forth, ask for its opinion. I promise at the end of that focused hour, you're going to come out with a whole new perspective on how to operate, whether it's personally, professionally, whatever.
I promise you're going to come out with a whole new perspective because the way it just thinks is so different from any other AI model you've ever used. The way it talks to you, it communicates. with you, it conveys its ideas to you, is so different, so unslop -like, so human -like, so intelligent, so smart, it will change your perspective on how you're working.
And I think you're going to get way better results. Now, I'm going to be honest. I think something scary is happening at Google.
And by scary, I mean like kind of exciting, kind of frightening. All these models came out within like four weeks of each other. Argon came out like a month ago.
Then carbon, which is an upgrade, came out right after that. And then neon came out right after that. Then hydrogen.
I think Google might be the first AI lab to hit recursive self -improvement. Their rate of release over the last couple of months has been significantly faster than the rate of release of any other AI lab. Astra 6 -2, Chad GBT's latest model, that took like two months.
to release since the last Astra. Fable 5 .5, it's been like two months since the last Fable came out, right? And so their rate of improvement all of a sudden has skyrocketed out of nowhere.
And this isn't just fake, like micro step improvement. This is real tangible. improvement that Google is putting out on a almost weekly basis at this point, they might have reached recursive self -improvement.
For those that don't know what a recursive self -improvement is, that is when the AI is improving itself. And every time you go through the loop of improve itself, release, improve itself, release, that goes faster and faster and faster and faster until it takes off or in the singularity, right? And so Google, just based on release schedule and the amount they're improving out of nowhere, they might be the first ones to recursive self -improvement.
And the first ones to recursive self -improvement are going to be the ones that win because then you can exponentially improve from there while all your competitors are still linear. That is kind of scary, kind of awesome, but hey, at least we're getting awesome products out of it. I know this video was a lot more chitter -chatter than me going through, you know, visual benchmarks and slideshows and 3D games and 3D simulations.
I think it was important I don't do all of that for this video because I think the Gemini strengths are so far outside of, you know, 3D pelicans on bikes. So, you know, let me know down below, are you cool with more just chatter -based videos? Because I want to convey ideas rather than just kind of going through benchmarks, which everyone else I'm playing.
goes through. Was that helpful to you? Let me know down below.
If you learned anything at all, leave a like. I really appreciate it. If you leave likes, it means the world.
Subscribe and turn on notifications. All I do is try to make helpful videos about my experiences with AI. I am so grateful you would choose to watch my content.
I just do what I love and then tell other people about it. So thank you, thank you, thank you for watching the content. I'll see you in the next video.
The Hook

The bait, then the rug-pull.

A flat declaration that Google's newest models beat Claude Fable and ChatGPT Astra, followed by a promise to clear up the names, the pricing tiers and which model to use.

Frameworks

Named ideas worth stealing.

00:53list

Gemini 4 tier map

  1. Carbon: flagship, Fable-class
  2. Neon: efficient middle, Opus-class
  3. Hydrogen: fast, Sonnet-class

A translation of Google's three model names into the Claude tiers readers already know.

Steal forchoosing which model to use per job
05:00concept

Backwards-looking vs forward-looking models

Most models return the average of the best past answers; the host argues Gemini 4 instead projects what will work next.

Steal fordeciding when to use a second model for planning
CTA Breakdown

How they asked for the click.

VERBAL ASK
19:50subscribe
“If you learned anything at all, leave a like. Subscribe and turn on notifications.”

Standard sign-off, plus a poll asking whether viewers want more talk-based videos instead of benchmark walkthroughs.

Storyboard

Visual structure at a glance.

claims list
hookclaims list00:00
tier map
valuetier map00:53
tank game demo
valuetank game demo10:30
Google One plans
valueGoogle One plans13:30
recursive improvement
ctarecursive improvement17:50
Frame Gallery

Visual moments.

One-click upgrade to your Google

Get more breakdowns in your search results

Add Modern Creator as a preferred source and Google shows you more of our breakdowns in Search, Top Stories, and AI Overviews. It only changes what you see, and you can undo it in your Google settings anytime.

Add to Preferred SourcesOpens your Google source preferences with us pre-loaded. Tick the box and you're done.
Watch next

More from this channel + related breakdowns.

13:43
Alex Finn · Review

The Hermes Mobile App is AWESOME

Alex Finn walks through the new Hermes Agent mobile app and sorts out which jobs belong to it and which still belong to other AI agents.

November 11th