Google has switched on Lyria 3.5 music generation inside the Gemini app for everyone. In a blog post dated 4 September, the company said the model is available to all users globally, on the web and in the mobile app. That is a notably wider release than most of what Gemini has shipped lately, though not an unlimited one.
What Lyria 3.5 does in the Gemini app
You describe the music you want and Gemini generates a track. Google says you can pick or describe a genre, and choose between vocal and instrumental styles. There are also new templates meant as starting points, covering things like background music and birthday tracks.
Length is now a choice too. Google says you can ask for a short track or a longer one, which was not the case when the model was confined to Google Flow Music. Lyria 3.5 landed there on 29 July, roughly five weeks before this rollout.
Google’s own pitch for the feature is deliberately small-scale. “Whether you need a custom backing track for a video, a unique brand jingle, or just a personalized ringtone, it’s easier than ever to bring your idea to life,” the company wrote in its announcement. Google also describes Lyria 3.5 as its best-sounding music model, which is its own claim rather than anything measured independently.
Available to everyone, but not unlimited
Two days ago Google put Gemini 3.8 Flash into the app for paying subscribers only. Lyria 3.5 goes the other way. The announcement names no AI Pro or AI Ultra requirement, and no region list either.
Read the small print before you plan around it. Google’s own support page for music generation says there are limits on how many tracks you can create with Lyria. Hit the limit and you get a notification, at which point your options are to upgrade your subscription or wait for the allowance to refresh. Google does not publish the number anywhere on that page.
So “available to all users” is accurate, and it is not the same thing as free use. The generous read is that Google wants everyone to try this. The practical read is that audio generation is expensive to run, and an unpublished quota is the lever that decides how much of it you get before the upgrade prompt appears. Two other requirements are worth knowing: you must be 18 or older, and Keep Activity has to be switched on.
The guardrails, and the watermark
Lyria’s safety filters block two specific things. One is a request for a particular artist’s voice. The other is the generation of copyrighted lyrics. Every track the model produces also carries a SynthID watermark, which Google describes as imperceptible. Google’s support page confirms the watermark is embedded in the delivered file, which arrives as a video with generated cover art.
That watermark is worth understanding correctly. It is designed to let Google’s own detection tools identify a track as AI-generated. It is not something you can see or check yourself. It also does not settle who owns the output. If you are making a jingle for a business, read the terms before you build anything commercial on top of it.
Where else it runs
Beyond the app, Google says Lyria 3.5 is available in Google Flow Music, in Google AI Studio, and in Google Vids. Developers get it through the Gemini API. Per Google’s API documentation, as reported by The Next Web, the API returns 44.1kHz stereo audio and accepts images as well as text prompts. Generation there is single-turn, meaning a track cannot be refined across several prompts.
One piece of context Google did not address. A Munich court ruled on 31 July that Suno had infringed copyright, in the first European judgment of its kind. That case turned partly on memorised works and infringing outputs. Lyria’s two named guardrails sit close to that territory. Google has not said the two are connected, and neither should anyone else.
Source: The Next Web
