In this article
- 1 Quick decision
- 2 What each one costs, verified
- 3 ChatGPT wins on the product around the model
- 4 Grok wins on velocity, agents, and price of entry
- 5 What I actually open Grok for, a month in
- 6 The honest budget wall
- 7 Does “buy ChatGPT first” still stand? Yes
- 8 Where this comes from
- 9 Pick your seat: four situations, four calls
- 10 Correction log
Quick answer: if you are picking one subscription today, pick ChatGPT. It is the more complete product and the $20 earns its keep every day. But Grok is not the punchline it was a year ago.
It went from a top-ten contender to a top-five model. xAI is out-building everyone on data centers, and Grok Bot just shipped the most interesting agent product of the summer.
ChatGPT earns today. Grok is a bet on next year. That’s my read, and the rest of this page is me showing my work.
Full disclosure up front, and it changed since the last version of this page: I’m still using it a month later. I’m paying on the hundred dollar plan, SuperGrok Plus.
The earlier version of this page said I had only used it on a trial and was trying to justify adding it. I added it.
So now I pay for both, and the verdict below didn’t change when the money did. That part surprised me.
Quick decision
| Your situation | Pick |
|---|---|
| First AI subscription, want the most product for $20 | ChatGPT Plus |
| On a budget, mostly writing and everyday questions | ChatGPT Go at $8, or Grok’s free tier |
| You live on X and want AI where you already are | Grok via X Premium at $8 |
| You want always-on agents doing jobs while you sleep | Grok Bot is the one to watch |
| You already pay for ChatGPT or Claude and have $30 spare | SuperGrok is the most interesting add-on of 2026 |
What each one costs, verified
I read these off the official pages on August 21, 2026, for my AI plan tracker:
- ChatGPT: Free at $0, Go at $8/month, Plus at $20, and two Pro tiers at $100 and $200. Per OpenAI’s own plan docs, Pro $100 is “5x higher usage than Plus” and Pro $200 is 20x. No annual billing on any of them.
- Grok: Free at $0 with “generous limits,” SuperGrok at $30/month, SuperGrok Plus at $100, per x.ai’s pricing page. The side door: X Premium at $8/month includes “increased usage limits on Grok,” and Premium+ at $40 goes higher.
xAI’s pricing page, captured August 21, 2026, SpaceX logo in the corner. Free, $30 and $100 tiers.
Prices on this page age. The tracker link above is where I keep them current, with a dated changelog.
ChatGPT wins on the product around the model
The product around the model. That’s the whole case for ChatGPT, and it’s a strong one.
The app is the best in the category, voice is the best in the category, and the feature list is deep: deep research, scheduled tasks, custom GPTs, Codex for anyone technical, and record mode. I use ChatGPT on my phone, on the go, constantly. It is the AI I talk to out loud.
The limits keep resetting in your favor. OpenAI has been unusually generous lately, resetting usage multiple times a month. When I hit a wall, the wall usually moves before I do. For a $20 product, the mileage is absurd.
It is the safe recommendation. If a stranger asks me “which AI should I pay for first,” I say ChatGPT without hesitation. Not Claude, not Grok. It’s just easier to use, and someone new gets more out of an all-rounder than a specialist tool.
Grok wins on velocity, agents, and price of entry
Velocity. This is what changed my mind about xAI. They’ve put more into data centers and hardware than anyone, and it shows in how fast the models improve.
Grok went from an afterthought to a top-five model in about a year. Betting on the team that ships fastest is a real strategy, and it is my bet: xAI is the SpaceX of this race.
Grok Bot. Launched August 11, 2026: always-on agents that get their own cloud computer, sign into your tools, finish multi-step jobs unsupervised, and coordinate with each other in group threads.
If you’ve wanted a self-hosted AI operator setup but couldn’t face the tinkering, this is that idea as a product. In my trial the multi-bot part was the thing that stuck with me: open a research agent, spin up a second bot, and watch them hand work to each other.
The X integration. Real-time search over X is Grok’s alone. No other assistant has that data, and for news, sentiment, and what-is-happening-right-now questions, it shows.
The bundle math. $8/month for X Premium with increased Grok limits is the cheapest way into a frontier model that exists right now.
What I actually open Grok for, a month in
The last version of this page couldn’t have this section, because I hadn’t paid for it yet.
Image generation, and specifically Grok Imagine. This is the real one. It is fast, it is good, and it gives you a spread of options instead of one image at a time. It reminds me of Midjourney when I did use it back in the day, and I do not pay for a dedicated image tool anymore.
What that opens up is prototyping. Imagine is where I plan to run thumbnail directions and visual brainstorming, and it is the tool I would reach for to mock up a website redesign concept for a client.
Seeing several directions fast changes how you think, and no other assistant I pay for gives me that.
My own Grok Imagine screen, September 10, 2026. One prompt, four takes. This is the spread I mean.
Grok Build is the other half. It’s actually in the running now. A year ago I wouldn’t have considered it at all.
It is not at ChatGPT’s frontier level, and I am not going to pretend otherwise. But it moved from “not worth opening” to “open daily,” and that is the fastest trajectory in this comparison.
My first Grok Build, September 10, 2026: a landscaping site from one sentence. Not client-ready, but it shows the direction.
The 60-second version, from my Shorts: Grok builds the app from one sentence and hands you the link.
So here’s what the $100 bought me: an image and prototyping lane I didn’t have, plus a front-row seat on the team shipping fastest. It didn’t replace anything.
The honest budget wall
The affiliate comparisons never write this part. I now pay for three of these, and each one has to earn its spot every month, not just impress me once.
Most readers face the same decision, just at a different scale. The question is never “which model is smarter.” It is “which subscription earns its spot this month.”
If your budget is one subscription: ChatGPT, and it is not close. If it is zero: Grok’s free tier plus ChatGPT’s free tier is a strong stack now.
Does “buy ChatGPT first” still stand? Yes
Paying for Grok didn’t move my recommendation. It would be easy to assume it did.
I still think ChatGPT is the one. With Astra I can feel the difference, and it is a very good model.
For me it comes down to usage more than benchmarks. OpenAI keeps resetting limits, and keeps being generous about it.
That’s a company that wants you to keep using the thing. Others are stingier with resets. I don’t expect it to last forever, but that’s where it is today, and it’s worth money.
Even without the resets, you get more working mileage per dollar. I feel like with Claude you hit it faster on the twenty dollar plan than you do with ChatGPT.
One caution on the $20 ChatGPT plan specifically: be careful with Astra there. It will eat your usage quickly. Use Sol at medium for everyday work and save the heavy model for the task that needs it. Same advice I give on Claude usage limits, for the same reason.
Where this comes from
Prices come from the official pages, read August 21, and live dated on the plan tracker. x.ai blocked my re-check on September 18, so the Grok tier prices are five weeks old. Open x.ai/pricing before you spend.
The experience calls are mine: a year-plus paying for ChatGPT, a month paying $100 for Grok. Here is the billing screen:
My own SuperGrok Plus billing screen. The Imagine and Build screenshots above are from the same account.
Pick your seat: four situations, four calls
- Newcomer with $20: ChatGPT Plus. Do not overthink it. If $20 stings, Go vs Plus is the $12 question.
- Power user with subscriptions already: keep what earns. If you generate a lot of images or prototype visuals, Grok is the add-on that justified itself for me.
- X native: Premium at $8 is the cheapest frontier-model seat in the game.
- Claude person wondering about the other side entirely: I wrote the Grok vs Claude version of this decision too, and the wider Claude vs ChatGPT call.
Correction log
- August 21, 2026. Published. I had used Grok on a trial only and said so.
- September 5, 2026. Corrected ChatGPT Go from $12 to $8, read off OpenAI’s rendered pricing page.
- September 18, 2026. The biggest correction on this page so far: I now pay for Grok, on the $100 SuperGrok Plus plan, a month in. Replaced the trial disclosure, replaced the “promo I missed” section with what I actually use it for, and added the current read on whether ChatGPT is still the one to buy first. It is. Noted that x.ai blocked price re-verification that day.
Written by
Chris AlarconChris Alarcon shows people who don’t code the boring Claude setup that actually works. He runs his one-person operation on these systems and shares every workflow, prompt, and tool combo in public. Start with the 15 prompts he uses every day.
The boring setup
It still won't know you tomorrow.
The Setup is one folder. Drop it into a Project, answer three questions, and Claude or ChatGPT writes your memory file and your voice file itself. It stops forgetting you, stops sounding like AI, and runs three real jobs by name. Twenty minutes tonight. No code. $39, 30-day refund.
See The Setup →The 15 AI prompts I actually use every day
I tested hundreds and kept the 15 that survived. You get them right away, then one short email from me every Tuesday.