What is Lyria?
Lyria is Google DeepMind's music generation model. The first version was announced on November 16, 2023 with YouTube's Dream Track experiment in Shorts. Lyria 2 became generally available on Vertex AI on May 20, 2025, and Lyria 3 came to the Gemini app as a beta on February 18, 2026, making 30-second tracks and writing lyrics on its own.
Google says all Lyria 3 tracks in the Gemini app carry SynthID, its hidden watermark for AI-made audio.
What's new in Lyria 3.5
Lyria 3.5 launched on July 29, 2026, first in Google Flow Music. Google lists richer melodic structure, better lyrics, more realistic vocals with better pronunciation, and easier control of tempo and length. On September 4, 2026 it reached the Gemini app, Google AI Studio, the Gemini API and Google Vids.
Google DeepMind's Lyria page describes songs of up to three minutes, and the Gemini API guide describes 44.1 kHz stereo audio with a SynthID watermark.
On MakeMidas, Lyria 3.5 writes a full song with vocals from your prompt, with your own lyrics or with lyrics it writes. It sets the length itself, and each song is one price.














