Lyria 3 Pro
Google DeepMind's flagship music generation model that composes full-length songs with structured sections, custom lyrics, and provenance watermarking from text or image prompts.
Overview
What is Lyria 3 Pro
Lyria 3 Pro is Google DeepMind's flagship music generation model, released in March 2026. It generates full songs up to three minutes long with structured sections like verses and choruses from text or image prompts, outputting high-fidelity stereo audio with built-in watermarking and content credentials.
Using it anonymously on Venice
On Venice, Lyria 3 Pro runs under an anonymized privacy tier — your prompts are not stored, profiled, or retained. You pay a flat $0.10 per generation with no subscription, giving you private, permissionless access to Google DeepMind's structured music model with zero retention.
Assessment
Strengths and limitations
- Full song generation with section-level structure (intro, verse, chorus, bridge) from text or image prompts.
- Vocal generation with custom lyrics and an instrumental mode for flexible creative control.
- Built-in audio watermarking and C2PA content credentials for output provenance and transparency.
- Detailed structure controls for duration, BPM, and intensity to constrain the composition.
- High-fidelity stereo MP3 output with global availability.
- Proprietary and closed: no open weights, self-hosting, or fine-tuning possible.
- No negative prompting support; you cannot exclude specific elements via negative prompts.
- Limited to one clip per prompt with a hard cap at 184 seconds.
- Active content filtering on inputs and outputs, including vocal likeness checks.
- Currently in Preview launch stage, so capabilities and availability may change.
Samples
Sample outputs
Generated on Venice with our standard prompt suite — the same prompts we run through every model of this type, so you can judge it like-for-like.
“An uplifting cinematic orchestral build with soaring strings, warm brass, and a hopeful resolution.”
“A mellow lo-fi hip-hop beat with a soft jazzy piano loop, vinyl crackle, and a relaxed late-night mood.”
Specifications
Datasheet
- Maker
- Google DeepMind
- Released
- March 25, 2026
- Launch stage
- Preview
- Modality
- Text-to-music, image-to-music
- Max duration
- 184 seconds (~3 minutes)
- Audio output
- MP3, 44.1 kHz, 192 kbps
- Languages
- English, German, Spanish, French, Hindi, Japanese, Korean, Portuguese
- Open weights
- No — proprietary
- Privacy on Venice
- Anonymized — prompts not stored
- Available on Venice since
- May 2026
API
Call it from your code
Venice exposes this model through the REST API. Queue a generation with the model id.
curl https://api.venice.ai/api/v1/audio/queue \
-H "Authorization: Bearer $VENICE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "lyria-3-pro",
"prompt": "An uplifting cinematic orchestral build with soaring strings"
}'
# Use the returned queue_id with https://api.venice.ai/api/v1/audio/retrieve.
# Call /audio/complete after downloading if needed.Pricing
What it costs on Venice
Flat per-image pricing on Venice: $0.10 per generation.
New Venice accounts include a free daily allowance and 500 welcome credits — no credit card required.
Alternatives
How it compares
| Model | Max duration | Best for | Open weights | Price (Venice) |
|---|---|---|---|---|
| Lyria 3 Pro | ~3 min | Structured songs & lyrics | No | $0.10 / track |
| ACE-Step 1.5 | — | Budget music generation | No | from $0.03 / track |
| MiniMax Music 2.6 | — | General music generation | No | $0.18 / track |
Google's flagship music model on Venice, offering section-level structure control, lyrics generation, and C2PA content credentials.
Use cases
What it is good for
- 01Prototyping full songs with defined structure for musicians, producers, and songwriters.
- 02Generating soundtracks and background scores for short-form video, ads, and games.
- 03Creating instrumental or vocal tracks from image-based mood and style references.
- 04Enterprise media workflows that require watermarking and verifiable content credentials.
Prompting
Getting better results
Describe song structure explicitly (e.g., 'intro, verse, chorus, bridge') for coherent composition.
Supply custom lyrics to guide the vocal generation style and narrative content.
Use image prompts to set the mood, genre, or atmosphere of the generated track.
Specify duration, BPM, or intensity to tighten control over the final output.
Version history
Consumer release with 30-second clip generation.
Current — extended duration, structured composition, and developer API release.
FAQ
Frequently asked questions
Lyria 3 Pro is Google DeepMind's flagship music generation model, released in March 2026. It generates full songs up to three minutes long from text or image prompts, with structured sections like verses and choruses, vocal and instrumental modes, and built-in watermarking.
On Venice, Lyria 3 Pro costs a flat $0.10 per generation. There is no subscription required; you pay per track with credits.
No, it is a paid model. You pay $0.10 per generation on Venice, with no subscription lock-in. New accounts may receive complimentary trial credits depending on Venice's current policies.
No. Lyria 3 Pro is a proprietary Google DeepMind model with closed weights. It cannot be self-hosted or fine-tuned.
Lyria 3 Pro excels at structured full-song composition with lyrics and section-level control, while MiniMax Music offers general track generation at a different price point. Choose Lyria for complex song architecture and MiniMax for lower-cost, shorter-form music.
Yes. You can provide an image to influence the style, mood, and genre of the generated music alongside a text description.
Venice runs Lyria 3 Pro under an anonymized privacy tier, meaning your prompts are not stored, profiled, or retained for training. No personal generation history is kept.
It outputs MP3 audio at 44.1 kHz and 192 kbps, with stereo support and a maximum length of 184 seconds per generation.
Use Lyria 3 Pro anonymously
Venice does not store your prompts. Chat history stays in your browser.