Lyria 3

Overview

Lyria 3 is a generative music and audio creation model from Google/Google DeepMind that was surfaced first through the Gemini app and later made available via API for developers. Based on newsletter mentions, it can transform text prompts or images into original music, including short 30-second clips and, in some product surfaces, longer full-length tracks. Its rollout positioned it as both a consumer-facing creative tool and a developer-accessible media model.

For AI Product Managers, Lyria 3 matters because it represents the productization of multimodal generative audio across both app and API channels. It shows how a foundation model can move from beta access in a first-party experience to broader developer adoption, while also incorporating provenance features like SynthID watermarking. This makes it a useful reference point for PMs evaluating AI media features, creator workflows, trust mechanisms, monetization limits, and API-based platform opportunities.

Key Developments

• 2026-02-19: Google DeepMind rolled out Lyria 3 in beta globally via the Gemini app for custom audio generation. Newsletter coverage noted that outputs include SynthID, Google's invisible watermark for AI-generated content.
• 2026-02-21: Google AI highlighted Lyria 3 alongside Gemini 3.1 Pro and other launches, describing it as a way to turn photos or text into dynamic music.
• 2026-04-10: Josh Woodward announced full access to Lyria 3 in the Gemini app, with up to 5 free full-length tracks per day, unlimited 30-second clips, and paid upgrades for additional music, images, and video generation.
• 2026-04-11: DeepLearning.AI described Lyria 3 as an AI music generator that transforms text prompts or images into original 30-second songs.
• 2026-05-26: Logan Kilpatrick announced that Lyria 3 is now available via the API, enabling developers to build products and workflows on top of the model.

Relevance to AI PMs

• Designing multimodal creative workflows: Lyria 3 is a concrete example of a model that accepts text and image inputs to generate music. PMs can use it as a benchmark for planning multimodal user journeys, prompt UX, and output controls for creative tools.
• Evaluating product packaging and monetization: The Gemini app rollout included free daily quotas, short-form versus full-length generation, and paid upgrades. This provides a practical template for thinking about usage caps, tiered plans, and cost management for generative media products.
• Planning trust and provenance features: The mention of SynthID watermarking is especially relevant for PMs building AI-generated media experiences. It highlights how provenance can be embedded directly into outputs to support transparency, policy compliance, and enterprise readiness.
• Transitioning from app feature to platform capability: Lyria 3 moved from Gemini app availability to API access. PMs can study this path when deciding whether a capability should remain a first-party feature, become a partner/developer platform, or support both.

Related

• Google / Google AI / Google DeepMind: The organizations most closely associated with launching and distributing Lyria 3.
• Gemini app / GeminiApp / gemini-app: The first-party surface where Lyria 3 was introduced in beta and later expanded with fuller access for users.
• API: A major milestone for Lyria 3 was becoming developer-accessible through API availability.
• SynthID: Google's invisible watermarking technology, referenced as being applied to every Lyria 3 output for AI-content provenance.
• Josh Woodward: Mentioned in connection with expanded Gemini app access and consumer packaging for Lyria 3.
• Logan Kilpatrick: Announced the API availability milestone, signaling the shift toward developer adoption.
• DeepLearning.AI: Covered the launch and described the model's text-and-image-to-music capabilities.