ChatGPT Gets GPT-Live: Two-Way Talk & Visual Cards

OpenAI is rolling out its new ChatGPT voice model, GPT-Live, to all users in stages. The biggest change is a two-way conversation feature, allowing users and ChatGPT to speak and listen simultaneously. Free users receive GPT-Live-1 Mini, while paid users get GPT-Live-1. You'll see a "Live" label on your mobile voice chat screen or web menu if it's applied. Users can also switch back to the previous voice model. This update allows for a more natural conversation flow. You can keep talking even if ChatGPT starts responding, and it will adjust its answer if you interrupt with a new question. There's also a new real-time translation feature. Users can now adjust the intelligence level of voice conversations, choosing between Instant, Medium, and High for different response styles. Visual cards have also been added, displaying graphic information on screen for questions like sports scores or weather. Early users report the features are stable and feel more like talking to a real person. However, some issues have been identified, including problems with ChatGPT memory in voice mode and English-style accents in foreign language pronunciation. OpenAI is actively investigating these. This update aims to transform ChatGPT's voice feature into a real-time interactive tool.