Android 17 Ships with Gemini Omni, Lyria 3, and a New Multitasking Layer
Google released Android 17 with a bubble bar for faster app switching, foldable gaming mode, and three new Gemini AI tools: Lyria 3 for music generation, Gemini Omni for in-conversation video editing, and AudioLM for real-time speech translation on Pixel 10a.
Android 17 is out, and the headline isn’t the redesigned multitasking system — it’s the three new Gemini AI models bundled into the Pixel Drop that shipped alongside it.
Google released the stable build on June 16. The OS update itself is centered on the new “bubble bar,” a persistent multitasking rail that replaces the recents screen for faster cross-app navigation. Foldable devices get a dedicated gaming mode. Parental controls gain finer-grained controls for app scheduling. Wear OS picks up a natural-language widget layer and Google claims up to 10% improvement in battery life for supported watches. Emergency detection, previously a Pixel exclusive, rolls to Wear OS.
The Gemini layer is the real story
The Pixel Drop accompanying Android 17 introduces three models that represent a significant jump in what Gemini can do on-device and in conversation.
Gemini Omni is a multimodal model with in-conversation video editing. You can share a video clip mid-conversation, ask Gemini to trim it, adjust color, add captions, or reframe the subject — and it handles the edits without leaving the chat interface. That’s a practical capability differentiation that ChatGPT doesn’t have in its mobile app today.
Lyria 3 is Google DeepMind’s updated music generation model, now accepting both text and images as inputs. Give it a photo of a foggy coastal landscape and a style prompt and it generates an original music track. The model ships integrated into Pixel devices and into the Gemini app. Generation quality, according to early reviewers, is noticeably better than Lyria 2 on rhythmic consistency and instrumental separation.
AudioLM brings real-time speech-to-speech translation to Pixel 10a. The model handles conversation in your language and delivers translated output at near-native latency, with an emphasis on preserving the speaker’s original tone and cadence rather than producing flat machine translation. It works offline for supported language pairs.
Android 17’s distribution advantage
This update demonstrates how Google’s platform ownership translates into AI distribution at a scale no startup can match. Android runs on roughly 3.4 billion active devices. Even if Gemini Omni, Lyria 3, and AudioLM only reach the Pixel install base initially, the fact that they ship integrated into the OS — not as a separate app download — changes the default user experience for hundreds of millions of people.
That’s the vector through which Gemini gained 7.7 percentage points of AI chatbot market share in the past year, per Sensor Tower data published yesterday. The next move for Google isn’t convincing users to switch — it’s making Gemini the obvious tool they already have open.
Wear OS gains
The Wear OS improvements are meaningful for developers targeting wearables. Natural-language personal widgets mean users can build device shortcuts by describing what they want (“show my next meeting and today’s steps”) without navigating a settings menu. The 10% battery improvement comes from system-level scheduling changes, not hardware. Both are in the stable Wear OS 6 build.
The stable Android 17 rollout begins immediately for Pixel devices, with OEM partners following in Q3 2026.