4.0 / 5
Good
- Free to install, with voice, photo input and image generation included
- Live voice conversation suits hands-busy moments and language practice
- Photo input covers useful jobs like transcribing handwritten notes
- Chat history syncs across devices, so you can pick up a conversation anywhere
Not so good
- Syncing means your conversations are stored with OpenAI, not just on the phone
- Answers are generated, so advice and facts need checking before you act
- On-screen help is only on limited release and needs deep access to your screen
- It drafts plans and copy but is not a place to keep or manage them
Ask it in plain words, get an answer back
ChatGPT is OpenAI's own Android app for its AI chatbot. You type or say what you want, in ordinary language, and it writes back: an explanation, a draft, a list, a plan.
That is the whole idea, and everything else in the app is a different way in. You can speak instead of typing. You can show it a photo. You can ask for a picture instead of text. The conversation stays the core.
It is also very widely used. Play puts it at over 1,000,000,000 installs, which tells you this is the mainstream option in AI chat rather than a niche one. Popularity is not quality, though, so the rest of this review is about what it actually offers.
For when your hands are busy: voice
Advanced Voice Mode turns the app into a spoken conversation. You tap the soundwave icon and talk, and the developer says it replies in real time, so it works more like a phone call than voice typing.
On a normal day that covers the walk to the station, the kitchen, or the car park. OpenAI pitches it for settling arguments and for language practice. The second one is the more interesting use: a patient partner who will happily go over the same phrase twenty times.
One caveat. A voice reply is harder to check than text on a screen. If the answer matters, read it back.
For pictures: reading them and making them
Photo input works in both directions. You can snap or upload a picture and ask about it. The developer's examples are transcribing a handwritten recipe and identifying a landmark, and both are good tests of the feature: turning messy real-world images into text or facts you can use.
Image generation is the other half. You describe a picture and it makes one, or you hand it an existing image and tell it what to change in words. For a birthday card, a quick mock-up or a joke for a group chat, that is a lot of capability in a free app.
It is not a photo editor in the usual sense. You get no sliders or layers, just instructions in plain language. People who want precise control over an image will still want a dedicated editing tool.
For work, study and the rest of the day
Most of what the app is for falls under general help, and OpenAI covers a lot of ground here. For work: brainstorming marketing copy or sketching out a business plan. For learning: explaining a topic at whatever level you ask for, child-friendly or refresher. At home: suggesting a recipe from whatever is left in the fridge, or putting together a travel itinerary.
It also offers to talk through personal problems and help you word a tricky reply. That is a reasonable use for a sounding board. But it is still a sounding board, not a counsellor.
The limit to keep in mind across all of this: it generates answers, it does not guarantee them. An itinerary might include a museum that is closed on Mondays. A business plan is a draft. Treat everything it writes as a first pass.
It is also a conversation, not a workspace. It can outline a budget or a presentation. The budget still ends up in Microsoft Excel and the slides still end up in Microsoft PowerPoint, or whatever you use for that.
For questions about your screen
There is one more input method, and it is the most invasive. Using Android's Accessibility Services, the app can look at what is on your screen and answer questions about it, so you skip the screenshot-and-upload step.
OpenAI says this is on limited release, so not everyone will get it. It is also opt-in. That is the right design, because letting any app read your whole screen is a big step. If you would rather it never sees more than you choose to send, leave it switched off.
What it costs you
The developer calls the app free, and voice, photo input and image generation all come with it. That is a generous starting point for anyone who just wants to try AI chat on their phone.
| Input | Good for |
|---|---|
| Typing | Drafts, plans, explanations |
| Voice | Hands-free questions, language practice |
| Photo | Transcribing notes, identifying places |
| Image request | Making or changing pictures |
| Screen share | Questions about what is on screen (limited release) |
The less visible cost is your data. Syncing history across devices means your conversations live on OpenAI's servers and are tied to you. Anyone who types personal details, work documents or health questions into it should be comfortable with that first.
The app is rated Teen. Parents handing it to younger children should know that.
Verdict
For a free app, the offer is broad. Typed chat, live voice, photo understanding and image generation together cover most of what people want from an AI assistant, and none of them is listed as a paid extra. That is why it scores well.
It loses marks on the terms. Your history is synced to OpenAI, the screen feature asks for access some people will rightly refuse, and everything it produces needs checking. People who want their data kept on the device, or who need answers they can trust without looking them up, should skip it.
For everyone else, it is a sensible first AI app to install. Just go in knowing what you are trading for it.
By Hafiz Rahman
Editor, tools and utilities ·
The app in this story
ChatGPT
com.openai.chatgpt