Introduction to ChatGPT-4o’s new features: voice conversations, real-time translation, and a multimodal assistant

ChatGPT-4o brings text, voice, and image understanding into a single conversation, and the day-to-day difference in how it feels to use is obvious: faster, more like communicating with a real person, and better suited for handling tasks you can “see and hear.” Below, through the most everyday scenarios, we’ll help you understand what exactly ChatGPT-4o has upgraded, and which settings are worth quickly adjusting.

Where ChatGPT-4o’s “all-around” upgrades are

At its core, ChatGPT-4o is multimodal: within the same conversation, you can send text while also describing your needs by voice, and you can upload images or files for it to read directly. Compared with the old workflow of “take a screenshot first, then type an explanation,” ChatGPT-4o is more like an assistant that can understand the materials right in front of it.

In addition, ChatGPT-4o’s conversational pacing is more natural—especially for tasks that require follow-up questions, added constraints, and rapid iteration—reducing the back-and-forth cost of confirmation. You’ll find it easier to treat ChatGPT-4o as a tool for ongoing collaboration rather than a one-off Q&A box.

Voice conversations and real-time translation: smoother cross-language communication

ChatGPT-4o’s voice conversations are closer to a “you say one sentence, it responds with one sentence” style of exchange, making it suitable for use while driving, walking, or when your hands are busy. For people who aren’t comfortable typing out what they want to say, ChatGPT-4o is also more friendly.

On translation, ChatGPT-4o supports quick switching among multiple languages, and you can have it provide instant, interpreter-style paraphrasing between two languages. A practical use case is: repeat what someone says in a foreign language during a meeting to ChatGPT-4o by voice, and it will immediately summarize the key points in your preferred language and provide sentences you can use to reply.

Image viewing, file reading, and on-screen assistance: treat it like a “hands-on tutor”

When you upload images, slides, or spreadsheets, ChatGPT-4o can explain, summarize, and rewrite directly based on the content—for example, turning a PPT page into a spoken script, or organizing a flowchart into an actionable step-by-step checklist. Combined with its data analysis capabilities, ChatGPT-4o can also distill information from files into verifiable key points.

In learning and debugging scenarios, ChatGPT-4o works well as a “tutor”: you screenshot the problem or paste the assignment requirements, and it will first break down the conditions, then provide solutions and practice suggestions tailored to your level. If you use the desktop app on a Mac, you can also summon ChatGPT-4o anytime with a shortcut (Option + Space), without frequently switching back to the browser and interrupting your flow.

Memory features and controls: more convenient personalization, and more control

ChatGPT’s memory feature will refer in later conversations to preferences you explicitly ask it to remember (such as writing style or work background), making ChatGPT-4o smoother to use over time. More importantly, this memory is manageable: you can see what it has remembered, delete individual items, or clear everything.

If you don’t want ChatGPT-4o to reference past information in conversations, you can turn off “saved memories” or “chat history” in settings, or use a temporary chat to avoid recording and updates. It’s recommended to store only “preferences that need long-term reuse” in memory, and keep sensitive information like account details, IDs, and health information confined to temporary chats.