Parker Prompts compares Claude, ChatGPT, Gemini, and Grok $20 AI plans through coding, writing, image/video, and research tests to find the best subscription.
Key Takeaways
- Claude Pro offers the best model placement across benchmarks, and produced the most polished coding result.
- ChatGPT Plus is the strongest all-rounder, leading on natural-sounding writing and realistic image generation.
- Grok ranked last in most tests but offers the best image editing tools and agent features at a higher price point.
- The best subscription depends on your primary use: coding and analysis, writing and images, or Google-app integration with storage.
What the video covers
- The video compares the $20-tier AI subscriptions of Claude Pro, ChatGPT Plus, Google AI Pro, and Grok (Premium at $30) on price, included models, and benchmark rankings.
- Artificial Analysis rankings show Gemini and Claude Opus 5.5 near the top, with Grok 4.7 well behind; the creator argues Claude Pro wins on model placement across seven rankings, though gaps are small.
- Coding test: each platform built a coffee shop website from one prompt. Claude (Sonnet 5.5) ranked first for animations and a complete, functional flow; ChatGPT second, Gemini third, Grok fourth.
- Writing test: the same brief produced a 7-minute YouTube script from each model. ChatGPT produced the most natural, human-sounding script; Claude second, Gemini third, Grok last for AI-like wording and giving away key numbers early.
- Image and video test: Claude has no image or video generation and is disqualified. ChatGPT produced the most realistic Instagram ad image; Grok offered the strongest image editing tools; Gemini and Grok were both tested for short video, with the creator wouldn't post either yet.
- Research/decision test: for a second-location decision, Claude and ChatGPT were compared for reliability and sourcing, with the creator finding Claude more reliable for this kind of analysis.
- Extra features discussed include computer-use/agent tasks (Claude in beta, Grok Bot with a cloud computer), scheduled tasks across platforms, and document/slide/design creation tools.
- Google AI Pro bundles Gemini into Docs, Sheets, and Slides, plus 5 TB of storage, though the in-app Gemini is less capable than the main model.
- Recommendation: pick based on use case — Claude for coding and reliable analysis, ChatGPT for writing and images and an all-around plan, Gemini for Google ecosystem and storage, Grok for image editing and agent-style tasks.
- The creator promotes a paid school community (AI Fluency) and provides free prompts and a scoreboard in the video description.
Chapters
- 00:00Introduction and Goal of the Comparison
- 00:56Plan Pricing and Included Models
- 01:58Benchmark Rankings Overview
- 02:45Coding Test: Coffee Shop Website
- 05:08Writing Test: YouTube Script
- 06:19Image and Video Generation Tests
- 08:10Research and Decision-Making Test
- 09:49Extra Features and Final Recommendations
Full Transcript — Download SRT & Markdown
Speaker A
If you're about to sign up for an AI subscription and you don't want to end up paying every month for the wrong one, you're in the right place because I ran Claude, Chad GBT, Gemini, and Grock through the exact same tests to see how
Speaker A
they hold up against each other. And by the end of this video, you'll know exactly which subscription to pay for based on how you use AI. But before we start, let's see how much each of those subscriptions costs, which AI models you
Speaker A
get, and which one gives you the best model for your money. Growth Pro is $20 a month and $17 if you pay for the whole year and you get Opus 5.5, Sonet 5.5 and Hiku 5.5. Their most advanced model
Speaker A
payable 5.2 isn't included unless you buy usage credits or move up to max. JGBT Plus is $20 a month with no yearly plan and it comes with a wider list of models including GBT6 Astra, 6.1 Soul and Terra alongside GBT6 Soul and Luna.
Speaker A
In the normal chat, you'll be using GBT6 soul and the rest are only in its work and codec tools. Google AI Pro is $19.99 and you get Gemini 4 argon, 3.8 Flash, and 3.1 Pro. Grog is the odd one here
Speaker A
because its closest plan is premium at $30 a month. For that, you get Grog 4.7 in Grog Build plus Cursor Pro and X Premium Plus. Now on artificial analysis, an independent ranking that test each model on stuff like coding,
Speaker A
science, and knowledge work. Gemini for carbon on high comes first with 60 points. Opus 5.5 is right behind it with 59. So 5.5 has 55 and Gro 4.7 is way down at 35. But on Arena where people vote between anonymous answers, Carbon
Speaker A
and Argon are on top with Opus 5.5 following them. So net 5.5 and Astra are almost tied further back and Grog 4.7 is slower still and across the five other rankings I checked first place keeps moving between CLA and JGBT. So if
Speaker A
you're comparing the models alone CL Pro wins here because it's the only plan whose best model comes first or second out of these four on all seven rankings though the gaps are pretty small. But these are just benchmarks not real world
Speaker A
results. So the first test I put them through is the most challenging one which is coding a full website with working advanced animations and fully functioning features all from just a single prompt. And all of these platforms come with their own dedicated
Speaker A
coding tools. So clo has clo code. JBT has codeex. Gemini has anti-gravity and gro has Gro built along with cursor.
Speaker A
I'll go ahead and ask each one to build a website for a madeup coffee shop with a menu, a quiz that picks your drink, and a cart you can actually check out from. Oh, and if you want to test this
Speaker A
out for yourself, I've left every prompt in this video along with a scoreboard on how these models perform down in the description for free. Anyway, here's what Cloode built me. For this, I used Entropic standard balanced model, sonet
Speaker A
5.5, with high thinking effort. And the reason I don't use Opus here is that it takes up too many tokens. And for coding builds like this, Sonet is pretty much on par with it. What got me right away
Speaker A
is the animations, like how each section fades in as you scroll down the page, and the buttons fill with color when you move your cursor over them. Everything else works too, and it's really easy to understand. So, it's a smooth experience
Speaker A
from the menu all the way to checkout. Next is JBT, where I used GBT 6.1 soul on high effort. On visuals and colors, this one immediately looks slightly cleaner than Cloth. It works and the overall UI is good, too, but it doesn't
Speaker A
have the fancy animations and functions that Claude gave me. It's still pretty good and usable website, though. Then there's Gemini, and it built this with Argon, the most advanced model you get on this plan. It also looks pretty nice
Speaker A
at first, but the spacing problems make it feel kind of weird, like how this text is right up against the edge of the screen. The UI is clean, but it's not as easy to understand as the others on this
Speaker A
test, and it's much flatter. And even though Argon scores higher than Sonet on both of those rankings, I'd say Sonet is the better one for the job like this, and it's more reliable. The last one is Grock. And again, the UA looks clean,
Speaker A
but honestly, it could be much better. It has spacing problems, too, and the cart options for the coffees are super flat with only text and nothing else. So CL takes the win in this round because it gave me the most finished website out
Speaker A
of all four and it executed the coffee shop website idea better. GBT comes second, Argon third, and Grog fourth.
Speaker A
Now, whichever one you end up going with. If you want step-by-step lessons on building things like this website with AI, that's exactly what my school community AI Fluency is for. You get a new video every single day for 12 weeks,
Speaker A
over a thousand done for you resources and direct help from me and my team inside the community. So, if you want to get in, the link to join is right down below. The second test is about writing, and this is where I want to see if any
Speaker A
of these models can give me a usable YouTube script out of the box. So, to do that, I'll fit each AI the same brief for a 7-minute video about that coffee shop. And I'll also include the facts it's allowed to use and nothing else.
Speaker A
All of them get the facts and math right. So the real difference here is how they sound and feel. And that's where Grock falls behind. It has the most AI like wording and a broken rhythm. It also gives away the major
Speaker A
numbers way too early. So I don't really have that many reasons to keep reading and that puts Grock in last place.
Speaker A
Gemini is a step up from Grock, but it still reads more like an article than a YouTube script that a person would say out loud. It even made up a random story about some customers that wasn't in my
Speaker A
brief and I specifically told it to stick to my facts. So, Gemini is third.
Speaker A
Cloth is better than both of them and the way it teaches stuff is very easy to follow and it explains things better than the other two. It does make some things up by itself and it still has a
Speaker A
lot of AI tells. So, it would need a ton of work and polishing before it's actually something I'd use which makes CL second. And Chad GBT is by far the best out of all of them because it has
Speaker A
the strongest hook with a human tone and rhythm throughout the whole script as if a normal person is talking. It stretches things out a little at the end, but with a few tweaks, I could genuinely use it.
Speaker A
So, TGBT is the clear winner of this round because it gives us the most natural sounding and usable script of all four right out of the box. And this next test is where it gets unfair because this time I'm testing them on
Speaker A
image and video generation. CL doesn't even have an image or a video generation model. So it's immediately disqualified.
Speaker A
Grock and Germany can do both, but JBT can only make images. So first I'll ask all three of them to create an Instagram ad for that coffee shop with a coffee back and a mark on a counter and the ad
Speaker A
text right on the image. Groat looks the most flat here. Gemini looks a lot better and it's pretty close, but JGBTS still looks the most realistic to me.
Speaker A
But when it comes to editing, Gro is by far the best because you get way more options like selecting individual segments or specific elements. JGBT has its own editing tools, too, like commenting on the image, removing the background, and marking it up. And
Speaker A
honestly, for most edits, that's all you need. And I'm only testing one ad here, but you can use these image models for all kinds of things like YouTube thumbnails, social media posts, product photos for an online store, and images
Speaker A
for your website or presentations. So if you make any kind of content or run a small business, this alone can save you a lot of time and money. Now for video, chip is out, too. So that leaves Gemini
Speaker A
and Groc. And I'll ask both of them for an 8-second ad of coffee pouring into a mug. I wouldn't post either of these because speed video generation just isn't that usable right now apart from maybe some little bureau shots that you
Speaker A
can't find online. So CHBT takes this round even without video because it gives you the most usable and realistic results and they actually look good instead of flat and lifeless. But not everyone is paying for this just to make
Speaker A
images. And the next test is where I want to see which one you can really trust with a big decision. So, let's say our coffee shop needs a second location in Portland. And before committing to a decision like that, I need proper
Speaker A
research I can rely on that tells me if this is a good idea or not. To do that, I'll give each AI a prompt telling it to go research things like the real costs and the local market and then give me a
Speaker A
verdict based on everything it finds. All four of them tell me to hold off for now. So, what I really care about is which one of them I can believe and which one actually helps me. Groc is the
Speaker A
easiest to rule out because even though it's mostly right, its answer is so generic that I can't decide anything from it. It just tells me what to do without much reasoning behind it. Gemini is a lot better and it suggests other
Speaker A
options like starting a coffee cart, but most of its numbers come from one company that sells coffee equipment, so I don't fully trust it. That leaves Clot and CHBT. And picking between these two is way harder. If I just want to know
Speaker A
what to do next, I'd go with Clot because it tells me exactly what has to happen before a second chop makes sense.
Speaker A
And the way it pulls all the information together is much more advanced than the other models. It's not perfect, though, because it writes way more than I asked for, and it even messes up its own budget total by $1,000. JGBT is the most
Speaker A
accurate one because it shows its math and tells me what it couldn't find, and its key numbers check out. So this round goes two ways because clo gives you the better analysis and helps you more with the decision. But JBT's information is
Speaker A
more reliable. Now beyond these tests, there are a lot of other perks and features that come with these AI subscriptions and they can make a big difference in which one you end up paying for. First, all of them have
Speaker A
agent features, which basically means the AI can go and do the work for you.
Speaker A
So it's doing more than just answering questions in the chat. You'll see this as co-work on cloth, work on JGBT, and tasks on Gemini where you point it at your files and have it handle the boring academy work. So, it can sort out a
Speaker A
massive folder, turn a pile of receipts into an expense report, or rename a bunch of files at once. And since that's mostly what it's doing, they're all pretty much on par here. CL and JGPT can even use your actual computer to do
Speaker A
tasks. So, they can click around your apps the way you would, though that part is still in beta on cloth. All of these platforms also support scheduling, which means you set up a task once, pick when it should run, like daily or weekly, and
Speaker A
the AI does it by itself at that time. So you could have it go through your inbox every morning and send you a summary or put together a report at the end of every week, and you never have to
Speaker A
ask again. Then there's integrations, which just means connecting your AI to other apps like Gmail, Google Calendar, Google Drive, and plenty more. Once an app is connected, your AI can see what's inside it and work with it for you. So,
Speaker A
it can pull information out and on a lot of these apps make changes right from the chat. But where they start to differ is the unique features each of them has.
Speaker A
Grog subscriptions includes Grock bot, an advanced AI agent with its own computer in the cloud where it can do real tasks like a human would and it keeps working even when you close your laptop. I've made more than a couple of
Speaker A
videos about it myself. So, you can check those out on my channel because it's really useful. There's also an automation page which is basically the same as scheduling, but it has readym made templates you can follow for things
Speaker A
like news and finance so you don't have to set everything up yourself. And your $30 plan gets you Cursor Pro, which is a coding tool, and X Premium Plus, which is the paid version of X. JBT's big extra is sites, which lets you create a
Speaker A
website inside the app, then host it and send people the link. For CL, it's something called artifacts. An artifact is a finished piece of work cloth makes, like a report or a slide deck that opens right beside the chat, so you can edit
Speaker A
it, come back to it later, and share it. And it has a whole suite of this with code docs for documents, code slides for presentations, a design canvas for things like mockups and dashboards for your numbers and charts, and for
Speaker A
spreadsheets, it makes real Excel files with working formulas. What makes this so useful is that you create stuff together side by side. So, Cloud writes the first version, then you type your own changes right into it or highlight a
Speaker A
part and ask it to fix that. You're not copying things back and forth between a chart and another app. So you get your work done a lot faster and better.
Speaker A
Gemini has something similar called Canvas, which is a workspace where it builds things for you, like a document, a website, or a full working app, and you can keep changing it until it's exactly what you want. You also get
Speaker A
Gemini built into Google's own apps like Docs, Sheets, and Slides. But the version in there isn't as smart as the main one, so it helps, just not that much, if I'm being honest. The best extra Gemini comes with is 5 TB of
Speaker A
Google Cloud Storage, which is a massive amount of space for your photos, videos, and files altogether. YouTube Premium Light is also included, so you see fewer ads and can download videos to watch offline. And there's Gemini Notebook, where you can load up to 300 sources and
Speaker A
ask questions about them. So, there is no single right answer here because the best AI subscription for you really depends on how you use AI. If you already use Google's apps for everything and you want a solid amount of AI built
Speaker A
into your daily workflow along with those 5 terabytes of cloud storage, go with Gemini. If you code a lot and you want to do your best work along AI, I'd recommend cloud because it's the strongest for building things and
Speaker A
thinking through big decisions. If you want a little bit of everything in one subscription and you don't need the cloud storage, JPT is the best fit because it's really good at writing. It makes realistic images and you can rely
Speaker A
on its research and it's still solid for coding. And I'd only recommend Grock if you specifically care about Grog bot or you want cursor and X premium plus along with it. Otherwise, I don't think it's really worth it for any task at this
Speaker A
price. And even though clo can't make videos on its own, I've got a full video on how to connect it to Higsfield and use it to design cinematic eye shots the way the top 1% do. To watch it, just
Speaker A
click the video on the screen right now. Thanks for watching and I'll see you in the next one.
Topics:AI subscriptionsClaude ProChatGPT PlusGoogle AI ProGrokAI comparisoncoding benchmarkAI image generationAI video generationbest AI plan 2026

![Central Cee – One Up [Music Video] — Transcript](https://i.ytimg.com/vi/mzpMpj0FHTU/maxresdefault.jpg)






![GW – My Gang [GANGSTER TRAP] — Transcript](https://i.ytimg.com/vi/vj9xMFAbRz8/maxresdefault.jpg)


