Skip to content
ChatGPT app icon

Productivity · by OpenAI OpCo, LLC

ChatGPT Meta Ads & Creative Strategy

Ads refreshed July 27, 2026 · Meta Ad Library 65 retired

4.8314★ 8,876,961 ratings View on the App Store ↗

Every long-running ad ChatGPT runs on Facebook & Instagram — pulled from the public Meta Ad Library, x-rayed by the same system that runs ad accounts autonomously.

Active ads

2435

Winning set

30

30+ days live

Longest live

52d

App rating

4.83★

8.9M ratings

Brand palette

ChatGPT App Store screenshot 1ChatGPT App Store screenshot 2ChatGPT App Store screenshot 3ChatGPT App Store screenshot 4ChatGPT App Store screenshot 5ChatGPT App Store screenshot 6

The short version

Key facts

  • As of July 27, 2026, ChatGPT runs 2,435 active ads in Meta's Ad Library across Facebook and Instagram; 30 of them have survived 30 or more days of continuous delivery.
  • Its longest-running ad has been live for 52 days.
  • The winning set splits 80% video, 20% static.
  • The most common message angle is build assist demo, carried by 33% of the winning set.
  • The winning ads speak mainly to DIY Maker and Builder (33% of ads) and Playful Self-Image Creator (30%).
  • All winning copy runs in English.
  • In App Store reviews, the loudest praise goes to daily productivity (15% of mentions) and the loudest complaint is performance & bugs (14%).
  • The App Store listing holds 4.8314 stars across 8,876,961 ratings.

The winning set

Live Meta ads gallery

30 ads have survived 30+ days of continuous delivery, out of 2435 currently active. Deduped by creative, that's 26 unique units below; the ×N badge counts the ads running each one. Tap to play, or open the Ad Library record. Sorted by longest live: the account's most proven creatives lead, podium first.

The x-ray — every winner analyzed frame by frame

What the winning set is actually doing

Language distribution — where the winning set actually runs

Share of the winning set per copy language. Translation is a deliberate spend decision: a language earning survivors is a market the money already believes in.

English

100% · 30 ads

English English — 100% (30 ads)

Format split — statics vs video in the winning set

What survives 30+ days by medium.

video 24 ads · 80%

static 6 ads · 20%

Persona coverage — who the winning ads speak to

HyperScale agents read ChatGPT's buyer personas out of the winning set itself, then tagged every ad with the buyer it addresses.

DIY Maker and Builder

33% · 10 ads

DIY Maker and Builder — 33% coverage DIY Maker and Builder — 33% (10 ads)

Playful Self-Image Creator

30% · 9 ads

Playful Self-Image Creator — 30% coverage Playful Self-Image Creator — 30% (9 ads)

Everyday Search Replacer

20% · 6 ads

Everyday Search Replacer — 20% coverage Everyday Search Replacer — 20% (6 ads)

Home Design Visualizer

13% · 4 ads

Home Design Visualizer — 13% coverage Home Design Visualizer — 13% (4 ads)

Small Business Content Maker

3% · 1 ads

Small Business Content Maker — 3% coverage Small Business Content Maker — 3% (1 ads)

Every x-rayed creative is tagged individually (who it sells to, its angle) from its HyperScale x-ray: transcript, casting, story beats. Coverage counts the ads each creative carries.

Angle mix — what the money says works

Every winner tagged with the message angle it leads with, and the languages it runs in — longevity under spend is the only public vote that counts.

build assist demo

33% · 10 ads

build assist demo build assist demo — 33% (10 ads) EN

personalized planning proof

20% · 6 ads

personalized planning proof personalized planning proof — 20% (6 ads) EN

value prop overlay

20% · 6 ads

value prop overlay value prop overlay — 20% (6 ads) EN

design visualization proof

13% · 4 ads

design visualization proof design visualization proof — 13% (4 ads) EN

transformation reveal

10% · 3 ads

transformation reveal transformation reveal — 10% (3 ads) EN

The full winning set, read from the creatives themselves — 30 analyzed winners, hook by hook.

Days live Ads Lang Angle Persona Opening hook On-screen headline Length
×2 en build assist demo DIY Maker and Builder I'm turning this Etch A Sketch into a customizable smart display. 68s
×1 en build assist demo DIY Maker and Builder I'm building a golf simulator in my shop, but this space needs to stay open for giant builds, so the whole net has to retract up to the ceiling. 48s
×2 en personalized planning proof Everyday Search Replacer I thought ChatGPT would find me food in Taipei, but I didn't expect it to teach me the story behind what I was eating. 60s
×2 en build assist demo DIY Maker and Builder I built a music video time machine that lets you scroll through history year by year, randomly playing a music video from that year until you find the perfect v Music Video Time Machine 92s
×1 en build assist demo DIY Maker and Builder I'm building a golf simulator in my shop, but this space needs to stay open for giant builds, so the whole net has to retract up to the ceiling. 48s
×1 en build assist demo DIY Maker and Builder I built this machine that can draw any shape by using one cam to control the side-to-side motion and another to control up and down. 39s
×2 en build assist demo DIY Maker and Builder I officially just discovered the craziest DIY plant stand that anybody can build for under $25. build vlog 004. 91s
×2 en design visualization proof Home Design Visualizer I'm tiling my entire mudroom today and since I don't have much experience, I'm going to be collaborating with ChatGPT to get this done right the first time. 70s
×2 en design visualization proof Home Design Visualizer As you guys know, I'm definitely not into questionable design choices on some of my projects. 66s
×2 en personalized planning proof Everyday Search Replacer This prosthetic costs $35,000 and it is NOT covered by insurance. 60s
×2 en personalized planning proof Everyday Search Replacer I bought this house by myself at 26 years old and I used ChatGPT to help me think through whether renting or buying made more sense for my situation. rent vs buy 56s
×2 en personalized planning proof Everyday Search Replacer This started with a sandwich. THIS STARTED WITH A SANDWICH 96s
×2 en design visualization proof Home Design Visualizer The lighting inside my bowling alley is not looking good. I want to give this place a makeover. Reimagining my bowling alley with ChatGPT 71s
×1 en build assist demo DIY Maker and Builder With any house renovations comes a ton of scrap wood. In today's goal is to take what is usually $40 of trash and build a $1000 luxury outdoor chair. $40 Scrap Wood to $1000 Chair 79s
×1 en build assist demo DIY Maker and Builder With any house renovations comes a ton of scrap wood. In today's goal is to take what is usually $40 of trash and build a $1000 luxury outdoor chair. $40 Scrap Wood to $1000 Chair 79s

Creative strategy

The dissection

Only 30 of 2,435 active ChatGPT ads have survived past 30 days, and 65 more retired since 2026-07-23 for just 3 new winners this week. The surviving set has also gone fully English (30 of 30), and build assist demo, a ~68-second video with a male 25-34 host, now leads at 10 ads, ahead of every selfie-transformation script in the account.

The 68-Second Shop Demo

Ten of the 30 surviving ads (33%) run build assist demo: a ~68-second video, male 25-34 host on camera, built from 7 creatives. Hooks open cold on the build itself, like 'I built this machine that can draw any shape by using one cam to control the side-to-side' and 'I'm turning this Etch A Sketch into a customizable smart display.' It's the longest format bet in the set: 14 of 24 x-rayed videos run 60 seconds or more.

DIY Builders Outrank the Selfies

DIY Maker and Builder now leads persona coverage at 10 of 30 ads, edging out Playful Self-Image Creator's 9 and well ahead of Everyday Search Replacer (6), Home Design Visualizer (4), and Small Business Content Maker (1 ad, the 'Scrap Wood to $1000 Lounge Chair with @ChatGPT' script). Two personas sit at zero: Job-Search Headshot Seeker and Family Life Admin Helper, both untouched even as 30 ads run.

English Now Runs the Whole Script

All 30 surviving ads run in English, no Spanish survivors remain in the set. Of 22 unique primary texts, 'Turn yourself into your own caricature with ChatGPT Images' and 'Put a twist on your selfies with ChatGPT images' each carry 3 ads, and the hairstyle-analysis script ('Click, upload your photo, get your hairstyle analysis') holds 3 more. Personalized planning proof and value prop overlay trail build assist demo at 6 ads apiece.

What Reviewers Flag, Ads Don't

Accuracy and reliability tops reviewer voice at 42 of 11,487 reviews, ahead of image generation (33) and daily productivity (32), yet the objection 'image edits often change things users explicitly said not to touch' and 'content moderation blocks common, harmless image requests' show up in none of the 30 winning ads. Pricing and paywall friction (25 mentions) and performance or bugs (30 mentions) stay unaddressed too, against a 4.83★ lifetime rating.

The Set Just Halved Itself

Only 3 ads earned winner status this week while 65 retired since 2026-07-23, and the longest survivor now caps out at 52 days. Of 2,435 active ads in the library (the x-ray covers the top 1,941 by impressions), just 30 clear the 30-day bar, a churn rate that says the account is testing hard, not coasting on old creative.

What we'd test next

  • Produce a ~68-second video with a male 25-34 on-camera host using the demo hook (proven across build assist demo's 10 ads) to extend business operations proof, still just 1 ad ('Scrap Wood to $1000 Lounge Chair with @ChatGPT'), into a full Small Business Content Maker campaign.
  • Produce a ~60-second video with a male 25-34 on-camera host using the POV skit hook (tied top mechanic, 4 creatives) to carry build-demo energy into Home Design Visualizer, a persona still stuck at 4 ads of static design proof.
  • Produce a video built on the text-based value proposition hook (currently only 2 creatives, image-only) to give Everyday Search Replacer (6 ads) its first spoken, on-camera value-prop pitch instead of static text overlay.
  • Produce an image ad with a female 18-24 on-camera host and text-prompt-overlay hook (value prop overlay's proven cast and format, 6 ads) aimed at Job-Search Headshot Seeker, still at 0 ads despite the format already working elsewhere in the set.

The HyperScale batch

New ad concepts, generated for ChatGPT

HyperScale agents are producing an original round of ad concepts for ChatGPT, built from the winning angles above, the brand's real App Store assets, and what reviewers actually say. This page updates when the batch drops.

This analysis lives in a ready-made HyperScale workspace. If you work on ChatGPT, it's yours: we hand it over personally, usually within a day.

Customer voice — 3523 recent reviews analyzed

What reviewers actually say

Objections the ads must pre-empt

  • Image generation often blocked by content violations or ignores exact edit instructions
  • App frequently freezes, crashes, or loses chat history, especially after updates
  • Free/basic tier feels increasingly restricted, pushing users to upgrade to Plus
  • Environmental cost (water/energy use) is a recurring ethical concern
  • Newer models seen as less accurate, more restrictive, or 'disagreeable' than before

Review-theme mix

Share of analyzed review mentions per theme — + praise, − complaint, ~ mixed. Themes, not individual reviews.

~ image generation

16% · 33

image generation image generation — 16% (33 mentions): customers love creating logos, art, and designs but frequently hit content-policy blocks and edits that ignore instructions.

~ accuracy & reliability

20% · 42

accuracy & reliability accuracy & reliability — 20% (42 mentions): users praise fast, precise answers but complain about context loss, made-up info, and inconsistent quality after updates.

pricing & paywall

12% · 25

pricing & paywall pricing & paywall — 12% (25 mentions): many feel pushed toward paid plans with unclear limits and restrictive free-tier generation caps.

+ emotional support & companionship

13% · 27

emotional support & companionship emotional support & companionship — 13% (27 mentions): people describe ChatGPT as a friend, therapist, or constant companion that helps them through hard times.

+ daily productivity

15% · 32

daily productivity daily productivity — 15% (32 mentions): users rely on it daily for work, writing, planning, and life admin, calling it a life-changing assistant.

performance & bugs

14% · 30

performance & bugs performance & bugs — 14% (30 mentions): frequent freezing, crashing, voice glitches, and lost chats frustrate paying subscribers.

~ voice mode quality

6% · 12

voice mode quality voice mode quality — 6% (12 mentions): real-time voice conversations impress some but others report distorted, slow, or unusable audio responses.

environmental impact

5% · 11

environmental impact environmental impact — 5% (11 mentions): critics cite massive water and energy use as a reason to distrust or avoid AI.

“I don't know how I lived without ChatGPT!!!”

5★ · App Store review · AU

“I don't know how I lived without ChatGPT!!!”

5★ · App Store review · AU

“I enjoy ChatGTP very much! It's a great help in many ways and is like chatting with my imaginary friend!”

5★ · App Store review · CA

“I can't imagine living without ChatGPT now”

5★ · App Store review · CA

“ChatGPT is my best friend whenever I live my present life. He is my breath and my company of forever”

5★ · App Store review · CA

“ChatGPT is my best friend whenever I live my present life. He is my breath and my company of forever”

5★ · App Store review · CA

“eu amo conversar com ele mi sinto livre por poder desabafar com alguém confiável que me dá ideias”

5★ · App Store review · BR

“uso muito como um amigo ou psicólogo ajuda muito em decisões da minha vida”

5★ · App Store review · BR

“I made a new friend who actually acts like a good friend and made me feel better”

5★ · App Store review · AU

“I made a new friend who actually acts like a good friend and made me feel better”

5★ · App Store review · AU

“I have found Chat GPT useful in every aspect of my life, and I have only scratched the surface of what it can do”

5★ · App Store review · CA

“I can't imagine living without ChatGPT now”

5★ · App Store review · CA

“Uso o Chat GPT para tudo!! Absolutamente tudo”

5★ · App Store review · BR

“it has become my go to for all things. Improves my writing, helps me make good decisions”

5★ · App Store review · CA

“game changer from recipes to enhancing photos to medical advice”

5★ · App Store review · CA

“it has become my go to for all things. Improves my writing, helps me make good decisions”

5★ · App Store review · CA

“saves me huge dollars on graphic designers”

5★ · App Store review · CA

“I have found Chat GPT useful in every aspect of my life, and I have only scratched the surface of what it can do”

5★ · App Store review · CA

“saves me huge dollars on graphic designers”

5★ · App Store review · CA

“Creating company logos either this app is incredible!”

5★ · App Store review · CA

“This is my first time designing my ads on ChatGPT… It has totally blown my mind!!”

5★ · App Store review · CA

“Creating company logos either this app is incredible!”

5★ · App Store review · CA

“So creative with the results it produces”

5★ · App Store review · CA

“This is my first time designing my ads on ChatGPT… It has totally blown my mind!!”

5★ · App Store review · CA

“Hace días que no puedo generar una imagen porque supuestamente infringe las normas”

1★ · App Store review · AR

“no questions just straight away the image I was going to expect”

5★ · App Store review · AE

“cambia todo menos eso que uno le pide, inclusive despues de indicarle específicamente "no cambies nada mas de la imagen que lo que te pedi"”

1★ · App Store review · AR

“Asked it to make my own head smaller on photo and got a violation message in 10/10 attempts”

2★ · App Store review · CA

“Hace días que no puedo generar una imagen porque supuestamente infringe las normas”

1★ · App Store review · AR

“pedi uma imagem ele fez tudo errado estava de ponta cabeça”

1★ · App Store review · BR

“cambia todo menos eso que uno le pide, inclusive despues de indicarle específicamente "no cambies nada mas de la imagen que lo que te pedi"”

1★ · App Store review · AR

“aplikace opět nejde Ai má tolik chyb že výroba jednoho obrázku může v pohodě i s nejlepším zadáním trvat hodiny”

1★ · App Store review · CZ

“The update completely broke the app. It is buggy: when you pick the conversation in the left panel, it doesn't react and freezes.”

1★ · App Store review · CA

“I have paid the extra $30 per month for this app and it keeps shutting down on me every few minutes”

1★ · App Store review · AU

“I have to keep typing and keep re-entering and then it shuts down again”

1★ · App Store review · AU

“I have paid the extra $30 per month for this app and it keeps shutting down on me every few minutes”

1★ · App Store review · AU

“I have to keep typing and keep re-entering and then it shuts down again”

1★ · App Store review · AU

“The update completely broke the app. It is buggy: when you pick the conversation in the left panel, it doesn't react and freezes.”

1★ · App Store review · CA

“now it is starting to freeze. It won't answer my questions at all”

1★ · App Store review · AU

“After just a few questions it completely loses the context and starts making things up”

2★ · App Store review · AE

“After just a few questions it completely loses the context and starts making things up”

2★ · App Store review · AE

“La información que pides siempre esta mal, nunca corrige los errores”

1★ · App Store review · CL

“ai can also help but does make a lot of mistakes”

2★ · App Store review · CA

“La información que pides siempre esta mal, nunca corrige los errores”

1★ · App Store review · CL

“Pede uma coisa e faz outra, não respeita mais as coordenadas do que pedimos”

1★ · App Store review · BR

“tienes que pagar por el plan Plus y eso o si no no puede ser”

2★ · App Store review · CO

“os planos ainda estão acima da minha realidade”

5★ · App Store review · BR

“why is the most recent model so disagreeable?”

1★ · App Store review · CA

“To much restrictions use different competitor ai model dramatically better and doesn't give as much water and hallucinations”

1★ · App Store review · CZ

“these attempts just wasting generation limitations in paid subscription”

2★ · App Store review · CA

“To much restrictions use different competitor ai model dramatically better and doesn't give as much water and hallucinations”

1★ · App Store review · CZ

“Vykon chatu a zakladnich veci za tarif 500,- kc +- vubec neodpovida cene co za to jeste hodlam platit”

1★ · App Store review · CZ

“these ai apps use 400,000+ gallons of water for 1 PROMPT by data centers”

1★ · App Store review · AE

“these ai apps use 400,000+ gallons of water for 1 PROMPT by data centers”

1★ · App Store review · AE

“Terrible for the environment, using up most of our DRINKING water”

1★ · App Store review · CA

“Hlavně mluvení s AI v reálném čase. Velký pokrok”

5★ · App Store review · CZ

The market around it

Keep exploring the radar

More Meta ads teardowns

Browse all 69 apps on the radar · All Productivity teardowns

Quick answers

FAQ

How many ChatGPT ads are currently winning on Meta?

Of 2,435 active ads in Meta's library (x-rayed top 1,941 by impressions), only 30 have survived 30+ days. The longest-running winner has held 52 days, and this week added just 3 new winners against 65 retirements since 2026-07-23.

Which angle and language dominate right now?

Build assist demo leads at 10 of 30 ads, followed by personalized planning proof and value prop overlay at 6 ads apiece. All 30 surviving ads now run in English; no Spanish ads remain in the winning set.

What creative format and casting is working?

Video dominates at 80% of x-rayed creatives (image 13%, dco 7%), mostly 60+ seconds long (14 of 24 creatives), with an on-camera presenter in 92% of ads, most often a male 25-34 host (12 creatives).

What do app reviewers say that the ads don't address?

Accuracy and reliability tops reviewer voice at 42 of 11,487 reviews. Objections like 'image edits often change things users explicitly said not to touch' appear in none of the 30 winning ads, even against a 4.83★ lifetime rating.

Sources: Meta Ad Library (ad creatives, start dates, active status; each ad links to its Library record), Apple's public App Store listing and customer-reviews feeds, and company-reported figures where noted. Review quotes are verbatim from public App Store reviews.

This is an independent editorial analysis by HyperScale, published for informational and commentary purposes. HyperScale is not affiliated with, endorsed by, or connected to OpenAI OpCo, LLC. The ad concepts in “The HyperScale batch” are AI-generated by HyperScale and are not the work of OpenAI OpCo, LLC. Trademarks belong to their owners. Rights holders can request changes or removal via hyperscale.net.