ChatGPT is OpenAI's flagship AI assistant — available on iOS, Android, Mac, Windows and the web — that handles everything from quick questions to long-form writing, real-time voice conversations, image analysis, file review, and live web search, all inside a single app. It's aimed at pretty much anyone: students, developers, writers, professionals who need a research shortcut. The current default model for paid users is GPT-5.6 Sol, a substantial jump from where this app started, and even free users now get GPT-5.5 Instant — a model that would have seemed implausible as a free offering just two years ago.
Key Features
- Real-time voice conversation — GPT-Live-1 (paid) and GPT-Live-1 mini (free) let you talk to ChatGPT naturally, with interruptions handled gracefully; it listens and speaks simultaneously rather than waiting for you to finish before processing.
- Live web search with cited sources — ChatGPT can pull current information from the web and show you exactly where each fact came from, which matters enormously when you're asking about something that actually changes over time.
- File and image uploads — Drop in PDFs, spreadsheets, code files, audio, or images (up to 512 MB per file, 10 files per message) and ChatGPT will read, summarize, compare, or analyse whatever you hand it.
- Memory and Projects — The rebuilt Dreaming V3 memory system (launched June 2026) stores facts across conversations automatically; Projects add a separate workspace layer with its own memory scope and uploaded files, so long-running work stays organised and doesn't bleed into your other chats.
Screenshots
Why You'll Love ChatGPT
Deep Research: the feature that actually justifies Plus
Deep Research isn't just a fancier web search. You give it a question, it spends several minutes running dozens of searches, reading sources, cross-referencing, and then produces a structured report with full citations. Plus subscribers get 10 Deep Research runs per month — enough to cover a serious research task every few days. If you regularly need to synthesise information across sources (competitive analysis, academic background reading, due diligence on a purchase), that alone covers the $20 monthly cost. Free users don't get it at all, and this is one of the most concrete reasons to upgrade if your work involves any kind of research.
Memory that actually sticks
The rebuilt Dreaming V3 memory system, which rolled out in June 2026, is a meaningful upgrade over what came before. ChatGPT now automatically extracts preferences, recurring context, and past decisions from your conversations and stores them in a profile you can view, edit, or delete at any time through Settings. Projects take this further by giving you isolated workspaces — each Project keeps its own memory and uploaded files separate from everything else, which means a work project and a personal project stop contaminating each other's context. For anyone who's been burned by ChatGPT "forgetting" important context mid-project, this fixes the core frustration.
Category: AI Assistants
Platform: iOS, Android, macOS, Web
Price: Free, with Go ($8/month), Plus ($20/month), Pro ($100 or $200/month), and Business ($20/user/month) plans
Frequently Asked Questions
Yes — the free tier is genuinely usable, not crippled. Free users get GPT-5.5 Instant as their default model, though with a message cap (roughly 10 messages per 5 hours on the flagship before it throttles or drops to a lighter model). US users also see ads on the free plan. If you need the full GPT-5.6 Sol flagship with no caps, no ads, Deep Research, Agent Mode, and Sora video generation, Plus costs $20/month. There's also a Go plan at $8/month that raises message limits but still shows ads and skips the premium features. For casual or occasional use, free is more than enough.
No. ChatGPT requires an internet connection for everything — the models run on OpenAI's servers, not on your device. You can view previous conversations while offline, but you can't generate new responses, use voice mode, or run web search without a connection. If offline AI is a hard requirement, you'd need a locally-run model like those available through Ollama or LM Studio instead.
The core app needs only an internet connection to function. Microphone access is required if you want to use voice mode, and camera or photo library access is needed if you want to send images or use the live camera feature. None of these are granted automatically — iOS and Android will prompt you individually before any of them activate, and you can revoke them at any time through your device's privacy settings without losing access to the text-based features.