Lyria 3 Pro

• 🎵 Google's Pro-tier music model generating full songs from text.
• 📏 Produces structured tracks up to about three minutes long.
• 🆕 Extends Lyria's short clips into complete, arranged compositions.
• 🌐 Supports vocals, lyrics, and multi-language generation across genres.
• 🔧 Prompt from a single text description for full-song output.
• 🏢 Available via Vertex AI alongside the base Lyria 3 model.
• 🎯 Aimed at content creators, producers, and application developers.
• 🔒 Built by Google with responsible-AI provenance in mind.

Google

Google is an American multinational technology corporation and one of the world's most valuable brands. A subsidiary of parent company Alphabet Inc., Google operates across search, cloud computing, consumer electronics, and artificial intelligence. Its…

Lyria 3 Pro is Google's Pro-tier music generation model, released in 2026, built to turn a single text prompt into a full-length, structured song. Per the catalog description, it generates complete tracks up to roughly three minutes long, with support for vocals, lyrics, and multi-language generation across a range of genres — moving beyond short prototype clips toward finished, arranged compositions.

Within the Lyria family, the clearest positioning is as the higher-capability tier alongside the base Lyria 3 release. Google introduced both Lyria 3 and Lyria 3 Pro on Vertex AI, with the Pro variant oriented toward longer, more complete song output. This places it among Google's broader 2026 generative lineup, which spans image, video, and text systems such as Nano Banana Pro, Veo 3.1 Full Quality, and Gemini 3.1 Pro Preview.

Google distributes Lyria 3 Pro through Vertex AI, positioning it for content creators, producers, and developers who want to build music generation into their own applications and workflows.

Because independent evaluations and detailed vendor benchmarks for this model are not available among trustworthy sources here, this description stays deliberately factual: it summarizes the model's stated purpose, output length, and multi-language, vocal-capable song generation rather than making performance or ranking claims.