Meta on Tuesday launched its first standalone Meta AI app, introducing a more personal voice-based assistant powered by Llama 4.
The app is designed to offer a conversational, voice-first AI experience, with features aimed at personalization and context awareness. Meta said the assistant “remembers context,” understands user “preferences,” and is built to feel more natural in interaction.
Meta AI app features
Built with Llama 4
Meta AI is powered by the Llama 4 model, aimed at delivering smarter, more helpful responses tailored to individual users.
Voice-first interaction
Users can speak naturally to Meta AI. The app includes a voice demo using full-duplex speech technology, which lets the AI generate spoken responses directly from conversational training, rather than reading written replies. This toggleable feature is available for testing, though Meta cautions users may face technical issues as it’s still in development.
Personalization and context
Meta said the AI improves as you keep using it. It can:
- Remember user preferences like travel interests or hobbies.
- Pick up details based on chat context.
- Use information shared through connected Facebook and Instagram accounts (via the Meta Accounts Center) to enhance responses.
- Draw on past interactions, liked content, and user profiles to deliver more relevant answers.
Image generation and editing
Users can create or modify images using voice or text commands within the app. These features are integrated across Meta AI tools.
Discover feed
The app includes a Discover section where people can view or remix AI prompts shared by others. Meta clarified that “nothing is shared” unless the user actively chooses to post it.
Cross-platform access
Meta AI is now integrated across all Meta platforms, including Facebook, Instagram, WhatsApp, Messenger, and Ray-Ban glasses.
Users can switch between glasses and the app in supported countries, picking up conversations where they left off via the history tab on app or web. However, transitioning from the app or web to the glasses mid-conversation is currently not supported.
Existing Meta View users will see their paired devices, settings, and media automatically transferred to the Meta AI app’s new Devices tab once updated.
Web experience and desktop tools
Meta AI is also rolling out updates to its web interface, optimized for large screens:
- Includes voice interactions and the Discover feed.
- Offers an upgraded image generation tool with controls for mood, style, lighting, and color.
- A document editor for generating text and images, with PDF export, is currently being tested in certain countries.
- Also testing document import features for AI analysis.
Control and settings
Voice input is optional and user-controlled. A setting called “Ready to talk” allows users to keep the voice feature on by default. An on-screen microphone icon signals when voice capture is active. Meta said it aims to improve the assistant over time based on user feedback.
Availability
The Meta AI app is available starting today on iOS, Android, and Web.
- Voice conversations, including the full-duplex demo, are launching in the US, Canada, Australia, and New Zealand.
- Personalized responses are currently available in the US and Canada.