ChatGPT beta code teases upcoming live camera feature

Remember when OpenAI blew everyone's minds six months ago by having ChatGPT talk to a dog in real time through a phone camera? That feature might finally be closer to reality.

According to Rowan Cheung on X, code discovered in the latest ChatGPT beta version 1.2024.317 reveals upcoming Live Camera video features. It looks like Advanced Voice Mode is about to get a pair of eyes.

Giving voice mode a set of eyes

The hidden code details four key features: live camera functionality, real-time video processing, visual recognition capabilities, and tight integration with voice mode.

Instead of snapping a photo and waiting for a text response, users will be able to point their phone at something and hold a continuous spoken conversation about what the model sees. The concept was first shown off back in May, when OpenAI demonstrated the voice mode reacting to a live camera feed of a dog.

Why it matters

We think this is the missing piece that turns ChatGPT from a clever voice assistant into an actual multimodal agent. Real-time visual understanding completely changes how people interact with AI in physical spaces, whether you are asking for repair advice while looking under a sink or asking for directions on the street.

That said, it is still just hidden code in a beta build, and OpenAI has not confirmed a release date. Processing live video streams with minimal latency is brutally expensive, so scaling this to millions of users without lag will likely be the company's biggest hurdle.