Google’s most powerful music AI, Lyria 3.5, integrated into Gemini: One-click generation of complete 3-minute songs
2026-09-05 08:00Models🔥 42.2 heat score
1sources
1days unfolding
42.2heat score
5mentions
SummaryAI generated
On September 5, 2026, Google officially launched the latest music generation model Lyria 3.5 in Gemini applications and APIs. This model can be invoked through Google AI Studio, Vertex AI, and Gemini API. Users can input text prompts to generate several-minute songs with complete structures, including verses and bridges (up to about 3 minutes). Additionally, there is the 30-second clip model Lyria 3 Clip for loop preview. All generated content is defaultly output in MP3 format, while Lyria 3.5 also supports WAV format export. This feature marks the transition of Google’s music AI from an internal experimental tool to a mainstream application. All generated content is uniformly scheduled through the Interactions API, and SynthID invisible watermark is embedded for copyright tracking and content traceability.
Google has launched the latest music generation model, Lyria 3.5, in the Gemini application and Gemini API, marking the transition of its music AI from an internal experimental tool to a mainstream solution. This model can be invoked through Google AI Studio, Vertex AI, and Gemini API, supporting the generation of MP3 or WAV format audio based on text prompts. Lyria 3.5 can generate complete songs of several minutes long, including multiple choruses and bridge sections, while Lyria 3 Clip generates 30-second clips for looping and previewing. Both models default to MP3 output, and Lyria 3.5 also supports WAV output. All generated content is uniformly scheduled through the Interactions API, and SynthID invisible watermark is embedded for copyright tracking and content traceability.