Google Launches Lyria 3.5 Music Generation Model in Flow Music

On July 29 2026, Google DeepMind released Lyria 3.5, its newest music‑generation model, into the Flow Music platform. The update brings measurable gains in musicality, lyrical coherence, vocal quality, and user‑controlled editing, according to a blog post published by Google Labs.

Flow Music is a web‑ and mobile‑based AI music platform that lets creators write, remix, and share full songs. The service is part of Google Labs, the company’s experimental AI initiative that has been used to preview products such as Gmail and Google Calendar. In Flow Music, Lyria 3.5 replaces the earlier Lyria 3 model and is the third major music‑generation release from DeepMind in five months.

Lyria 3.5 is described as a system that can synthesize high‑quality audio from a text prompt. The model’s improvements target three core areas that are critical for music production:

Musicality – The model now produces more natural harmonic progressions and better maintains musical continuity across verses and choruses. Lyrics – Text generation is more coherent, reducing the random word placement that has plagued earlier AI‑generated songs. * Vocals – Vocal synthesis shows a noticeable leap in naturalness and emotional expression.

In addition to these audio‑quality upgrades, Flow Music offers granular creative controls. Users can adjust tempo, key, instrumentation, and vocal style directly in the interface, allowing the model to serve as a rapid prototyping tool rather than a replacement for human musicians.

The launch places Google in direct competition with other generative‑audio players. OpenAI’s Jukebox, released in 2020, remains one of the most well‑known AI music systems, though the company has been relatively quiet on new audio releases since then. Stability AI continues to iterate on its own music models, and the industry is watching how Google’s new model stacks against these competitors.

According to the Google blog, the improvements were achieved through a combination of larger training datasets and refined neural‑network architectures. The post notes that Lyria 3.5 can generate full songs with vocals, instrumentation, and production effects in a single pass, a feature that has been a goal for AI music tools for several years.

For creators, the update means that AI‑generated tracks are closer to studio‑quality production. Producers can use Flow Music to generate hooks, chord progressions, or full arrangements and then refine them in a digital audio workstation if desired. The platform’s royalty‑free licensing model allows users to publish and share songs without additional clearance steps.

Industry analysts see the release as part of Google’s broader push into generative AI across media. DeepMind has also launched text‑to‑image model Imagen, text‑to‑video model Veo, and the Gemini family of large‑language models. Lyria 3.5 adds music to that portfolio.

Google’s announcement did not include pricing details for Flow Music. The platform offers a free tier with limited credits and a premium tier that unlocks additional features, including higher‑resolution audio and extended editing tools.

The model’s release is the latest milestone in a series of AI‑driven music tools that aim to streamline the creative process. While the technology is still evolving, the improvements in vocal quality and lyrical coherence suggest that AI‑generated music will become increasingly viable for commercial use.

Google has not announced plans for a commercial product beyond Flow Music, but the platform’s public availability indicates that the company is testing the market response to a fully integrated AI music creation workflow.

In summary, Lyria 3.5’s entry into Flow Music marks a significant step for Google in the generative‑audio arena. The model’s enhanced musicality, lyrics, and vocals, combined with user‑controlled editing, provide creators with a powerful new tool for rapid music production. The move also intensifies competition with OpenAI and Stability AI, as the industry watches how these technologies will shape the future of music creation.