reviews.jev
ChatGPT reviews, read one by one
The 100 newest App Store reviews of ChatGPT: what the text says next to what the stars say, what people complain about, and who blames an update.
try: MyFitnessPal · Howbout · Storm Radar · MarineTraffic · Baby Tracker · Opal · ChatGPT
ChatGPT
Productivity · 4.8★ store rating
in words
What 100 reviews of ChatGPT say
The 100 newest reviews average 4.3 stars, and 81 of them give four or five. Read as text, 72 are positive, 5 mixed, 20 negative and 3 neutral. 8 of the four- and five-star reviews carry a complaint in the text.
0 of 100 reviewers describe an unexpected charge, a misleading trial or trouble cancelling (Productivity apps average 3.2%). 1 say the problem started with an update (category average 4.6%). 2 say they have left or are about to.
Complaint topics
reviews out of 100; one review can name several topics
Reviews worth reading
The ones with the most concrete detail, by Jev's score.
★★★★☆ · 2026-09-18 · v1.2026.244
Limits
The app is really good in itself. It talks back well and has a good memory. But I don’t really like how there are hourly and daily limits on chats. I think the whole experience would be made way better if there wasn’t.
★★★★★ · 2026-09-18 · v1.2026.244
ChatGPT Has Become My Everyday Sidekick
I absolutely LOVE ChatGPT! I honestly use this app every single day, and it has become one of those things I can’t imagine not having anymore. I use it for everything — researching things I want to know more about, shopping and comparing products, planning trips, helping me write emails and messages, keeping track of information, troubleshooting problems, and sometimes just having a good conversation. What I love most is that I don’t have to know exactly how to word a questi...
★★★★☆ · 2026-09-19 · v1.2026.251
Great graphics / questionable research
I love ChatGPT for visualizations, and I think it’s quite good at writing. But in terms of research, find a lot of its conclusions questionable, flat out wrong, or invented out of thin air.
★★★☆☆ · 2026-09-18 · v1.2024.348
Could do with some improvements.
I really only use this app for generating images because before when I asked it a question I’d remember the response and it’s give me a completely different response. When generating images I counted you can generate 6 images a day which is reasonable but you shouldn’t have to pay money to get more images. There are other ai’s out there that I generate as many images as I please.
★★☆☆☆ · 2026-09-19 · v1.2026.251
Accuracy and reliability decline
Accuracy reliability and clarity and even responsive had significantly declined over the last couple of months. Sometimes I get responses with a lot of embedded code I get a lot of apologies because of ChatGPT not taken into account my request ignoring some of my request or disregard some of my requests or just providing responses that don’t align. It’s very disappointing and simile more and more desirable just to not use it
These are reviewers' own claims, labelled by a model, and nobody verified them. 100 reviews is a small sample, so a gap of a few reviews against the category average means nothing. Found a mistake? [email protected].
New reviews every day, with keyword rankings and chart positions next to them.
how it works
A star rating is one number. The text has the reasons.
Every tile is one review. First the tiles are coloured by stars. Then each one is recoloured by what its text says, and the tiles that change are the reviews a star average hides: four stars with a complaint inside, five stars typed over an angry paragraph. In our study of 1.1 million reviews only 26% of four-star reviews were plain praise.
After that the tiles sort themselves by what the review is for (praise, a bug report, a complaint about a product decision, a billing problem, a feature request), and the last view lights up what stands out for this app, biggest first: its most common complaints, surprise charges, blame for an update, people saying they are leaving, and reviews that are waiting for a reply. The category averages next to the charge and update counts come from the same study.
Frequently asked questions
- How does the review analyzer work?
- It takes the 100 newest App Store reviews of the app you pick and sends the text of each one to Jev, a model from TypeSafe AI that answers fixed questions with a choice or a probability. Jev never sees the star rating, so the stars and the text can be compared.
- Is it free?
- Yes. Opening an analysis that already exists is always free. Each visitor can also send a few new apps to be analysed per day.
- How accurate are the labels?
- In a hand check of the surprise-charge label on the 1.1 million review study, about 9 in 10 flagged reviews were correct. Short or sarcastic reviews are the usual misses. 100 reviews is a small sample, so treat small differences as noise.
- Which reviews are used?
- The newest 100 reviews AppSigma has collected for the app across App Store storefronts, in any language.