Title card reading 'I Pay for Four AI Assistants. Here's the One I'm Canceling' with Claude, Gemini, Perplexity, and ChatGPT labels

I Pay for Four AI Assistants Every Day. Here’s the One I’m Canceling — and the One Thing Holding Me Back

I pay for four AI assistants. Every month, real money leaves my account for ChatGPT, Claude, Gemini, and Perplexity. I use all four. I have opinions about all four. And I have decided I am done paying for one of them.

This is not a spec-sheet roundup. It is what actually happens when one person runs all four chat assistants side by side for real work — a blog, my wife’s real estate business, daily lookups, the constant background research my job demands — and has to decide which subscription survives the next billing cycle.

Here is where I landed, and the one feature that almost talked me out of it.

The assistantWhat I payWhat it does for meVerdict
Claude$100/mo (Max, full price)My daily driver. Home, blog, everything that matters.Keep. No question.
Gemini$10/mo (Verizon perk)Fast second seat. Quick searches and lookups.Keep. Cheap and getting better.
Perplexity$0 (free year, grandfathered)Answer-engine search with sources.Keep while it’s free.
ChatGPT$20/mo (full price)Image generation. Staging real estate photos.Cancel — with one caveat.

Claude: the one I actually live in

Claude is my daily driver and it is not close. It is the tool I open first at home, the one running this blog, and the one I trust with the work I cannot afford to have done sloppily. I pay full price for it — $100 a month on the Max plan — and unlike the others I have no discount, no carrier perk, no expiring promo softening the blow. I just pay it, because it earns it.

Worth being honest about the sticker, though. Claude Pro is $20 a month. I am not on Pro. I burned through Pro’s limits in about fifteen minutes of real work and spent the rest of the day waiting for a reset, so I went to Max at $100. That is five times the price of the entry plan, and there is a $200 tier above me I have not needed. If you are deciding, know that the useful tier of Claude is the expensive tier. For me it is the single most valuable software subscription I have. For someone doing lighter work, Pro at $20 may be all they ever touch.

The useful tier of Claude is the expensive tier. I pay it because it earns it.

Gemini: the fast second seat for ten bucks

Gemini is my second tool, and it is on this list partly because of a deal that is almost unfair. Google AI Pro runs $19.99 a month at retail. I pay $10 for it because Verizon offers Google’s AI plan as a perk on myPlan and myHome accounts — half off, attached to a bill I was already paying. If you are a Verizon customer and you are paying full freight for Gemini somewhere else, go fix that today.

I do not use Gemini the way I use Claude. I reach for it when I need a fast answer or a quick lookup and do not want to spin up a real working session. It is quick, it is tied into Google’s stack, and at ten dollars it does not have to be my favorite to be worth keeping. It just has to be useful and cheap, and it is both.

Perplexity: the free one I would be foolish to cancel

Perplexity is the easiest call on the board: it costs me nothing. I got a full year of Pro free through one of the Venmo/PayPal promotions that floated around in 2025. Pro is normally $20 a month, so this is a real $240 of value I am riding out at zero cost.

One honest note for anyone reading this hoping to copy me: that specific Venmo/PayPal deal closed at the end of 2025, so you cannot grab it now. But Perplexity keeps running these — student plans, carrier partnerships, referral stacks. If you want an answer-engine search tool with real citations, watch for the next promo instead of paying $20 out of pocket. While mine is free, it stays. When the free year ends, it goes back on the chopping block with everything else.

ChatGPT: the one getting cut

Which brings me to the one I am canceling. I pay $20 a month for ChatGPT Plus — the same $20 it has cost since the day it went paid, years ago. And when I look honestly at what I use it for, the answer is almost entirely one thing: image generation.

I do not run my day-to-day chats through ChatGPT anymore. Claude does that. I do not use it for fast lookups. Gemini does that. I do not use it for sourced search. Perplexity does that. Strip those away and what is left is a $20 subscription that exists to make pictures. For a single feature, that is a hard line item to defend — especially sitting in a stack that already costs me $110 a month before ChatGPT even shows up.

Strip away everything the other three do better, and ChatGPT is a $20 subscription that exists to make pictures.

The caveat: that image generator is genuinely good

Here is the part that made this harder than the spreadsheet says it should be. ChatGPT’s image generation is excellent, and I have built a real workflow on top of it. I use it constantly to stage photos for my wife’s real estate business — taking a room and showing it furnished, warmed up, presentable. It works, I am fast at it, and I have gotten comfortable. Comfort is its own kind of lock-in. The honest reason I had not already canceled was not that the math was unclear. It was that I did not want to rebuild a workflow that already works.

That is exactly the trap worth naming out loud: paying twenty dollars a month, indefinitely, to avoid an afternoon of relearning. Said plainly, it sounds like what it is.

The test that actually decides it

So before I cancel, I am running one test — and so should you if you are in the same spot. The question is whether Gemini can replace ChatGPT for my staging work. Because if it can, the decision makes itself: I am already paying for Gemini at ten dollars, and dropping ChatGPT saves me twenty.

The encouraging news is that the image gap has flipped. As of 2026, the consensus from hands-on comparisons is that Google’s latest image model — the one everyone calls Nano Banana — has pulled ahead of ChatGPT on raw image generation, scoring higher on realism and visual quality, and returning results in seconds where ChatGPT’s newest model can take a minute or two to “think” through an image. ChatGPT still holds an edge in one place: precise, instruction-heavy editing — changing specific things in an existing image with exact text and tight control.

For real estate staging, that distinction matters. A lot of my work is generative — show this room warm and furnished — which now favors Gemini. Some of it is fussy editing, which may still favor ChatGPT. So the test is concrete: take five real listing photos, stage each one in Gemini, and see if the output clears the bar I have been hitting in ChatGPT. If it does, ChatGPT is gone and I keep the twenty dollars. If Gemini falls short on the editing-heavy shots, I have a real reason to keep paying — and now it is a reason, not a habit.

Keep a tool for a reason, not out of habit. The way you tell the difference is you run the test.

The money math

Laid out plainly, here is the monthly stack: Claude at $100, ChatGPT at $20, Gemini at $10, Perplexity at $0. That is $130 a month on chat assistants alone. Cutting ChatGPT takes it to $110 — about $240 a year back in my pocket for a feature I have a strong shot at replacing with a tool I am already buying.

Two hundred and forty dollars is not life-changing money. But “it is only twenty bucks” is exactly the logic that leaves you paying for four overlapping subscriptions and wondering where the money went. The discipline is not in any single cut. It is in being willing to actually look.

What I am doing — and what you should

Keep Claude if it is your real working tool. Pay for the tier you actually need, and do not pretend the $20 plan is the one doing the heavy lifting if it is not.

Keep Gemini if you can get it cheap — and if you are a Verizon customer, get it cheap. At $10 it does not need to win anything to be worth it.

Keep Perplexity while it is free, and do not pay retail for it — wait for the next promo.

Cancel ChatGPT if, like me, you have quietly migrated everything else off it and it survives on one feature. But run the replacement test first. Do not cancel on principle and then discover three weeks later that the thing you cut was carrying real work. Prove the replacement, then pull the plug.

I am running my five-photo Gemini test this week. My strong expectation is that ChatGPT comes off the bill by the end of the month. I will report back on whether Gemini actually held up for the staging work — because the test is the whole point, and “I think it’ll be fine” is not a test.


Reviewed, and opinion by Brad Rowland — IT Infrastructure and Operations leader, automation builder, and AI implementer. All four assistants paid for and used daily through June 2026.

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top