Google Releases Lyria 3.5 Music Model in Gemini Chatbot
Google has released its newest music model, Lyria 3.5, inside the Gemini chatbot and made it available to developers through its API on September 4, 2026.
The model is described as Google's 'best-sounding music generation model' with improvements in expressive vocals and richer arrangements.
Lyria 3.5 can be used by anyone with a Google account to generate full-length songs based on text prompts, including genre, instruments, beats per minute, and mood.
The model also supports multimodal input, allowing users to upload up to ten images to accompany the text prompt.
Google has included several features in the Gemini app to make music generation more accessible, including a template system for creating background music or custom birthday tracks, adjustable track length, and a user interface for selecting genre and vocal/instrumental styles.
The model is also available to developers through the Gemini API, which allows them to integrate music generation into their own products.
However, Google has included several limitations on the use of Lyria 3.5, including single-turn music generation (meaning iterative refinement of a generated clip across multiple prompts is not supported) and safety filters that block prompts requesting specific artist voices or copyrighted lyrics.