Google has released Lyria 3.5, the newest version of its AI music generation model, directly inside the Gemini app and through its developer API, the company announced this weekend. The rollout puts text-to-music generation in front of Gemini's enormous user base and gives developers programmatic access to the model for the first time in this generation.
The launch was reported by The Decoder and other outlets, which noted that Google is positioning Lyria 3.5 as its most capable music model yet, with better vocals and fuller arrangements than its predecessor. It is also the latest sign that the AI tools race now runs straight through creative media — something we follow closely in our AI tools coverage.
What Lyria 3.5 Can Do
According to Google, Lyria 3.5 delivers more expressive vocals and richer musical arrangements than the version it replaces. Users can pick a genre and style in the Gemini app, choose between vocals and instrumentals, and generate tracks that range from short clips to longer pieces of music.
Google has also added templates designed to help newcomers get started, including presets for background music and personalized birthday songs — a sign that the company is aiming at casual users and celebrations as much as at musicians. The pitch is familiar from Google's other Gemini features: lower the barrier until anyone can produce a finished artifact from a sentence or two of description.
The workflow reflects that simplicity. A user describes the kind of track they want, selects genre and style parameters, and decides whether the output should carry vocals or stay instrumental. Because the model handles both short stings and longer compositions, the same tool covers a birthday message, a podcast bed or a soundtrack for a home video — use cases that previously required either a stock-music subscription or a hired composer.
Where the Model Is Available
The rollout goes beyond the Gemini app. According to Google's announcement, Lyria 3.5 is also available in Google Flow Music with added features, in Google AI Studio for developers, and in Google Vids, the company's AI-powered video creation tool. API access means third-party applications can build on the model directly, which could seed a wave of music-powered features in third-party apps in the coming months.
That breadth matters. Embedding a music model across a consumer assistant, a developer platform and a video editor gives Google multiple surfaces where generated audio becomes a default creative option rather than a standalone novelty.
For developers, AI Studio access is the notable piece. It moves Lyria from a consumer curiosity to a building block: fitness apps could generate adaptive workout music, game studios could score levels procedurally, and education tools could produce custom songs for lessons. Google has not announced API pricing in the materials reported so far, so how the model is metered remains to be seen — but availability alone signals that the company wants Lyria 3.5 embedded in the broader app economy, not confined to its own products.
The Licensed-Training Pitch
The most pointed part of Google's announcement concerns training data. Google says the Lyria model was trained only on licensed content — an explicit contrast with Suno, the AI music startup whose training practices have drawn copyright litigation, including a landmark German lawsuit brought by music rights group GEMA that Suno lost earlier this year.
Google has not shared specifics about what training data it actually used, as The Decoder noted. But the licensed-content framing draws a clear line in an industry where the provenance of training data is becoming a legal and commercial differentiator. For rights holders, a major lab claiming fully licensed training raises the bar for competitors; for business users, it reduces at least some of the legal risk that has made AI-generated audio a fraught asset in advertising and media production.
A Crowded, Litigated Market
AI music generation is becoming one of the more contested corners of generative media. Suno's German court defeat over copyright questions signaled that the legal environment for models trained on protected recordings is hardening, particularly in Europe. Against that backdrop, Google's decision to emphasize licensing — while keeping the details confidential — reads as both a legal shield and a marketing message aimed at cautious enterprise customers.
The competitive stakes are also about distribution. Gemini's user base gives Lyria 3.5 reach that standalone music startups struggle to match, and integration into Google Vids means AI soundtracks could become a standard part of workplace video creation almost overnight.
What It Means for Creators
For working musicians, the arrival of a polished, licensed-data music model inside a mainstream assistant is a double-edged development. It lowers the cost of jingles, backing tracks and personalized songs to nearly zero, while raising familiar questions about how human artists compete with instant, adequate output. For now, Google's messaging centers on assistance and celebration-scale creativity rather than replacing studio production.
What is clear is that AI music is moving out of the demo phase. With Lyria 3.5 living inside the Gemini app, Flow Music, AI Studio, Google Vids and a public API, generating a usable song is now a feature, not a product category of its own — and the industry will be watching whether the licensed-training claim holds up to scrutiny.
Stay Ahead of AI
The AI tool landscape changes daily. Get the latest AI developments in one place.
Read more AI news →
