Lyria 3 lands in Gemini

This is less about novelty than distribution: Google is taking a research-grade music model and dropping it into a mainstream consumer AI surface. That matters because multimodal AI gets real when creation tools live inside the app people already use, not in isolated demos.

• –The biggest shift is product surface area: Gemini now spans text, images, video, and music generation in one workflow
• –The 30-second cap keeps Lyria 3 lightweight and social-first, which makes it more useful for quick prototyping than full music production
• –Support for prompts from images and videos makes this a genuinely multimodal feature, not just another text-to-audio box
• –SynthID watermarking gives Google a clearer safety and provenance story than many generative media rivals
• –For developers, this signals where consumer AI products are heading: bundled creative modalities instead of separate single-purpose tools