Google Lyria 3.5 Brings Full-Length AI Music Generation to Gemini API

Google has expanded its AI music generation offering with Lyria 3.5, an upgrade that brings full-length song creation to the Gemini API in public preview. The model can generate 44.1 kHz stereo music from text and image prompts, while Google also rolled it out to Flow Music with controls for elements such as tempo and duration. The change gives developers and creative teams a clearer route to build music generation into their own products and production workflows.

The most important difference from the previous Lyria 3 experience is scope. Google had made Lyria 3 available in the Gemini app for 30-second tracks from text or image prompts. Lyria 3.5 moves beyond short-form generation with longer outputs, while Google says the model improves musicality, lyrics, vocals, and creative control. Google's Gemini API changelog lists Lyria 3.5 as a public-preview model as of September 3, 2026.

What Lyria 3.5 changes for music generation

Lyria 3.5 is being deployed across two related Google surfaces. Flow Music received the upgrade on July 29, 2026, where Google described a prompt-driven music workflow with tempo and duration controls. The Gemini API documentation subsequently established the developer route, placing the model in public preview for applications that need to generate music programmatically.

For teams producing video, audio, or interactive experiences, the value is not simply the ability to make a short musical idea. Full-length generation can make AI-created music more relevant to longer videos, podcast segments, branded audio concepts, and product experiences where a brief clip may not be sufficient. Results still need creative review before publication, particularly where pacing, lyrics, and brand fit matter.

Google's documented capabilities include:

• Full-length song generation at 44.1 kHz stereo through Lyria 3.5.
• Text and image prompts as inputs.
• Improved musical coherence, lyric generation, vocal output, and creative control.
• Flow Music controls that include tempo and duration prompting.
• SynthID watermarking for AI-generated audio.

SynthID is significant because it provides a provenance marker for AI-generated content. It does not settle every practical issue around how a track can be used, but it makes the origin of generated audio more identifiable within Google's system.

From short clips to longer tracks

The comparison does not mean every Lyria 3.5 result will be ready to use without editing. Google has announced stronger output quality and longer-form generation, but the supplied materials do not define performance benchmarks, specific commercial usage terms, or a fixed level of access for every Google surface.

Practical workflow opportunities and limits

The Gemini API path is the key development for companies that want music generation to be part of a repeatable process rather than a standalone creative experiment. A developer could connect prompt creation, image selection, track generation, human review, and asset delivery inside an existing application or production pipeline.

Potential use cases include creating draft background music for marketing videos, developing audio variations around a campaign concept, or generating initial music options for podcast intros and other audio assets. Image prompting may also help a team connect a visual campaign direction with an audio brief. These are workflow possibilities supported by the model's documented inputs and API availability, not a promise that generated tracks will automatically meet a particular brand or production standard.

Pricing remains an important open question. The supplied research confirms that pricing documentation exists for Gemini API and enterprise paths, but it does not provide verified Lyria 3.5 prices or a complete breakdown of Flow Music access tiers. Organizations considering implementation should therefore check current Google documentation before estimating volume costs or committing to a workflow.

For businesses exploring generative audio, the practical task is to identify where music generation reduces repetitive early-stage work without removing editorial judgment. Lyria 3.5's longer outputs and API access make that assessment more worthwhile than it was for a short-clip-only workflow.

If your team wants to turn generative media capabilities into reliable business processes, Scalevise's AI workflow automation service can help map the right use cases, connect AI tools to existing systems, and retain human review where it matters. A practical implementation can reduce manual production steps while keeping control over quality and approvals. Request a consultation to discuss an AI automation project.

Frequently Asked Questions

What is Google Lyria 3.5?

Lyria 3.5 is Google's upgraded AI music generation model. It supports full-length song generation at 44.1 kHz stereo and accepts text and image prompts through the Gemini API public preview.

Where is Lyria 3.5 available?

Google rolled out Lyria 3.5 in Flow Music on July 29, 2026. It was also documented in public preview through the Gemini API changelog on September 3, 2026.

How does Lyria 3.5 differ from Lyria 3?

Lyria 3 was available in the Gemini app for 30-second tracks. Lyria 3.5 adds full-length song generation and Google describes improvements in musicality, lyrics, vocals, and creative control.

Can developers use Lyria 3.5 in their own applications?

Yes. Lyria 3.5 is available in public preview through the Gemini API, giving developers a route to integrate text- and image-prompted music generation into applications and workflows.

What does SynthID do for Lyria 3.5 audio?

Google applies SynthID watermarking to AI-generated audio. It is designed to help identify content as AI-generated.

Conclusion

Lyria 3.5 turns Google's music generation offering into a more capable longer-form audio tool, with Flow Music controls for creators and a public-preview Gemini API for developers. Its 44.1 kHz stereo output, multimodal prompts, and SynthID watermarking establish a stronger foundation for practical creative workflows, although organizations still need to confirm current access and pricing details before planning production use.