Google has launched Guided Vision in Gemini Live on compatible Android devices, a feature that delivers real-time audio descriptions of whatever the phone's camera is pointed at.

The rollout began today. Users who share their camera feed within Gemini Live can have Google's AI perform several concrete tasks: reading small text, describing the surrounding environment, finding or identifying objects nearby, and describing details on specific objects. The company announced the feature on its official blog.

The most immediate use case is reading fine print. Point the camera at dense contractual text, medication labels, or product packaging, and the AI narrates the content aloud in real time. Beyond text, the system is built to interpret a full scene — describing a room, locating a misplaced item, or calling out visual characteristics of an object the user is holding.

Google positions Guided Vision squarely as an accessibility tool. The company says it is designed for people who are blind, have low vision, or want assistance in specific situations — a framing that puts the feature in the same category as Apple's VoiceOver Live Recognition, which Apple has added to the iPhone and the Vision Pro.

The parallel with Apple is direct. Both companies are racing to turn on-device generative AI into a real-time visual assistant: continuous camera input, continuous audio output, no manual photo capture required. Apple's version lives inside its long-established VoiceOver accessibility stack; Google's lives inside Gemini Live, the company's conversational AI interface for Android.

According to Google, Guided Vision is available not only through the Gemini app but also more broadly, with availability extending beyond the core app experience as the company described in its announcement.

The feature requires a compatible Android device. Google did not specify the full list of supported hardware or any regional restrictions in the announcement.

For Google, Guided Vision represents a concrete, shipping application of multimodal AI — models that combine camera input and audio output in a live session — rather than a demonstration or roadmap item. Accessibility has become one of the most commercially and reputationally significant proving grounds for this class of models, because the utility is measurable: text read correctly, objects identified accurately, scenes described usefully.

The competitive dynamic now centers on which platform — Gemini Live on Android or VoiceOver Live Recognition on iPhone and Vision Pro — can deliver the most reliable real-time visual description. Both features are live. The differentiation will come down to accuracy, latency, and the breadth of devices each company can reach, and Google's next steps on device compatibility will determine how widely Guided Vision is adopted.