Overview
Lyria 3, released February 18, 2026, is Google DeepMind’s most advanced AI music generation model and the first to ship inside the Gemini app. It targets everyday users and creators who want to turn ideas—or a single photo—into a 30-second track with vocals and lyrics, without writing a single note. Compared to Lyria 2 (Vertex AI, instrumental-only), v3 adds image-to-music, auto-generated lyrics and vocals, and lets you specify style, vocals, and tempo in prompts. Tracks are exportable at high quality and tagged with SynthID for AI content detection. For a broader comparison of options, see our Best AI Music Generators roundup.
What's New
Image-to-Music and Multimodal Prompts
Lyria 2 accepted only text. Lyria 3 adds image and video as inputs: upload a photo or clip and describe the mood or scene; the model composes a track whose lyrics and feel align with the visual. This makes it easy to score a moment, a memory, or a mood board directly inside Gemini.
Auto-Generated Lyrics and Vocals
Lyria 2 on Vertex AI was instrumental-only and did not generate lyrics. Lyria 3 generates full vocals and composes lyrics from your prompt, so you can request a “birthday tune” or “song about a rainy day” without writing lines yourself. You can still describe vocal style in prompts—e.g. gender, range, timbre, and tone—for more control over the final sound.
Coherent Structure and Export Quality
The model produces 30-second tracks with clearer flow from start to finish and more coherent sections. Output is suitable for sharing or light production use. Every track carries SynthID watermarking so that AI-generated content can be detected even after edits or re-encoding.
Availability in the Gemini App
Lyria 3 is the first Lyria version built into a consumer app: it runs inside Gemini (desktop first, then mobile), with no separate API or Vertex setup. You access it by asking Gemini to create music in supported languages (e.g. English, German, Spanish, French, Hindi, Japanese, Korean, Portuguese).
Availability & Access
- Where: Lyria 3 is available in the Gemini app (web and desktop first; mobile rollout follows). Vertex AI docs currently list Lyria 2 (lyria-002); Lyria 3 is announced for Gemini (and Dream Track).
- Age: You must be 18 or older to use music generation in Gemini.
- Regions: Available in all countries where the Gemini app is offered; exact rollout may vary by region and platform.
- Languages: Music generation is available in 8 languages (e.g. English, German, Spanish, French, Hindi, Japanese, Korean, Portuguese); lyrics compatibility may vary.
Pricing & Plans
- Free tier: Available to all users 18+ in the Gemini app; limits apply. Google AI Plus, Pro, and Ultra subscribers get higher limits (exact caps not published and may change).
- Billing: No separate Lyria subscription or per-track fee; access is tied to the Gemini app and its plans (e.g. monthly or annual).
Best For
- Creators who want a fast soundtrack for Shorts, stories, or mood reels from a photo or idea
- Anyone experimenting with AI music without learning DAWs or writing lyrics
- Marketers and social teams needing short, on-brand music for campaigns or templates
- Multilingual projects that need lyrics and vocals in several languages
- Users who prefer a single app (Gemini) for both chat and music generation
FAQ
How is Lyria 3 different from Lyria 2?
Lyria 2 runs on Vertex AI, is text-only, instrumental-only, and aimed at developers. Lyria 3 runs in the Gemini app, adds image (and video) input, auto-generates lyrics and vocals, and is aimed at consumers. Both produce high-fidelity 30-second-style output; v3 adds multimodal input and full vocal generation.
Do I need a paid plan to use Lyria 3?
No. Lyria 3 is available to all users 18+ in the Gemini app; limits apply. Google AI Plus, Pro, and Ultra subscribers get higher music-generation limits; exact numbers are not published.
Can I use Lyria 3 for commercial projects?
Gemini’s terms and generative AI use policy apply to Lyria 3 output. Tracks are watermarked with SynthID for identification. Check Google’s terms and responsible AI use policy for commercial use, attribution, and restrictions before using output in paid or commercial work.
What languages does Lyria 3 support?
Music generation is available in 8 languages (e.g. English, German, Spanish, French, Hindi, Japanese, Korean, Portuguese). Lyrics support and coverage may vary by language; the Gemini app’s language list may expand over time.
Where can I try Lyria 3?
Use the Gemini app (gemini.google.com or the Gemini mobile app) and ask it to create music. You can describe a track in text or upload an image and describe the mood; Lyria 3 will generate a 30-second track with vocals and lyrics. No separate sign-up for Lyria is required beyond having access to Gemini.
