An AI music agent built by Google Labs and Google DeepMind. It began as the open-source music generation project Riffusion, grew into Producer.ai, was acquired by Google in February 2026, and relaunched in April of the same year under the name “Google Flow Music.” The old domain producer.ai now redirects to the current site at flowmusic.app. At its core sits DeepMind’s Lyria 3 family of music generation models: describe the song you want in a chat, and lyrics, composition, vocals, and mixing all follow in one continuous flow. Music video generation with Veo, stem separation, and audio effects live on the same screen, so you can turn an idea into a finished track without ever having touched an instrument or a DAW.
Key Features
- Songs completed from chat instructions alone: Describe something like “a bittersweet city pop track for late summer, female vocals,” and Gemini — which orchestrates the conversation — interprets the intent and returns a song complete with lyrics, arrangement, and vocals. Anything you dislike can be revised by continuing the conversation
- Full-length, vocal-capable generation with Lyria 3: Expanded well beyond the short previews of early versions, it generates full-length tracks running several minutes. Vocals are supported, and the successor model Lyria 3.5 adds more expressive singing, more natural arrangements, BPM control, and stem export for full-length songs
- Partial remixing with Replace and Extend: Replace swaps out a specific section via a text prompt without regenerating the whole track, while Extend continues a song past its endpoint while preserving its style. “I only want to change the chorus” no longer means starting over
- Stem separation and audio effects: Generated songs can be split into parts such as vocals and drums and exported. Useful if you want to finish the track yourself in a DAW, or if you need an instrumental version
- Music video generation with Veo: You can create visuals matched to your song inside the same service. An image model handles album art as well, so audio, cover, and video all come from one place
- A free tier to start with: Sign up and you get a credit-based free allowance to try generation. Subscribers to Google’s AI plan are told they can use it at no additional cost
Pricing
Pricing is credit-based, with higher tiers allowing more songs per month.
| Plan | Overview | Approximate volume |
|---|---|---|
| Free | Available with sign-up only; daily credits are granted | On the order of tens of songs per day |
| Starter | Entry-level paid plan | On the order of hundreds of songs per month |
| Plus | Mid tier | Several times Starter |
| Member | Top tier | Several times Plus |
Pricing information is current as of August 2026. The official pricing page could not be retrieved automatically, so please check the exact amounts and credit allowances for each plan on the official pricing page.
Pros & Cons
✅ Pros
- You can produce songs with vocals from plain-language instructions alone, with no knowledge of music theory or DAWs
- Generation, partial revision, music video, and sharing are all completed inside one service, with no tool switching
- Replace and Extend let you rework just one part while keeping the sections you like
- Stem export means AI output can be brought into your own production as a starting point
- A free tier lets you verify quality before paying
⚠️ Cons
- Because it is credit-based, a free allowance runs out quickly once you start iterating
- The acquisition and rename changed both the service name and the domain, so older articles and bookmarks often conflict with the current state
- Fine-grained mixing and mastering still fall short of dedicated DAWs and plugins
- Commercial use rights depend on your plan and the terms of service, so review the terms before using output for business purposes
- Pronunciation and intonation in non-English lyrics can sound less natural than in English
Comparison with Similar Services
| Criteria | Google Flow Music (formerly Producer.ai) | Suno | Udio | Stable Audio |
|---|---|---|---|---|
| Provider | Suno | Uncharted Labs | Stability AI | |
| Core model | Lyria 3 family (DeepMind) | In-house model | In-house model | In-house model |
| Vocal generation | Supported | Supported | Supported | Mainly instrumental |
| Partial editing | Replace / Extend | Supported (extend, cover, etc.) | Supported (remix, etc.) | Limited |
| Video generation | Music videos via Veo | Not supported | Not supported | Not supported |
| Free tier | Yes | Yes | Yes | Yes |
For audio alone, Suno and Udio are also highly polished, but the ability to produce a music video inside the same service is a clear differentiator for Flow Music. For production assets such as sound effects and loops, Stable Audio is the better fit.
Who Is It For
- Creators who want original music for videos or games, prepared themselves with simple rights handling
- People who have never touched an instrument or a DAW but want to hear the song in their head
- Social media publishers who need a song and a music video as a set, on short notice
- Existing track makers who want AI to handle the first draft, then export stems and finish in their own DAW
- Anyone already subscribed to Google’s AI plan looking to add tools available at no extra cost
Summary
Producer.ai has carried the same core experience — a song emerging from a chat — through its journey from Riffusion to Producer.ai to Google Flow Music. Full-length vocal generation with Lyria 3, partial editing via Replace and Extend, stem separation, and Veo-powered music videos together cover music production from entrance to exit on a single screen. Start with the free tier to see how far your ideas translate into sound, then consider a paid plan once volume becomes the constraint. And because the name and domain have changed, begin any research from the current official site.
Related Links
- Official site (Google Flow Music)
- Pricing
- Documentation
- Getting started
- Official blog
- X (formerly Twitter)
- Discord community
Pages on the old
producer.aidomain redirect to the current site (the pricing page redirect was confirmed as of August 2026).