New AI models on Soundverse: Google Lyria 3.5, Veo 3.1, Seedance 2.5 and more
Lyria 3.5, ElevenMusic v2.5, Veo 3.1, Seedance 2.5, Seedance 2.5 Edit and Gemini Omni Flash 1.1 are live on Soundverse. See them tested on one Halloween brief, and how to use them.
Quick answer: Six new models are live on Soundverse. For songs: Google Lyria 3.5 and ElevenMusic v2.5. For video: Veo 3.1, Seedance 2.5, Seedance 2.5 Edit and Gemini Omni Flash 1.1. To test them, the team gave each one a Halloween brief and showed exactly what was asked and exactly what came back, side by side, with no cherry-picking. You pick the model from the model dropdown in the song, music, video or music video generator, or just ask Agent One to choose for you.
New AI models on Soundverse: Google Lyria 3.5, Veo 3.1, Seedance 2.5 and more, tested on a Halloween brief.
The new models at a glance
| Type | Model | Where you'll find it |
|---|---|---|
| Song | Google Lyria 3.5 | AI Song Generator, AI Music Generator |
| Song | ElevenMusic v2.5 | AI Song Generator, AI Music Generator |
| Video | Veo 3.1 | AI Video Generator, AI Music Video Generator |
| Video | Seedance 2.5 | AI Video Generator, AI Music Video Generator |
| Video | Seedance 2.5 Edit | Editing existing footage |
| Video | Gemini Omni Flash 1.1 | AI Video Generator |
They sit alongside Soundverse's own music models, including Soundverse v9.
Why more models matter
Every model has its own strengths. Some are better at certain genres, vocals or moods. Some video models are better at motion, some at realism, and some at editing footage you already have. On Soundverse you don't have to pick one platform and live with its limits: the leading models are in one place, so you can choose the right one for each job, compare results, and keep working without switching apps.
The Halloween test
The video runs one Halloween brief per model and shows the result as it came back. A few examples from the video:
- Lyria 3.5: Trick or Treat Queen, Halloween electro-pop at 125 BPM in G minor.
- Veo 3.1, with its own sound: a dog in a witch's hat that says "I'm not trick-or-treating. I am the treat," then sneezes twice.
- Gemini Omni Flash 1.1: a whimsical 3D animation of a ghost and a cat, for 960 tokens.
The point isn't that one model wins. It's that you can see what each one does with the same kind of brief, then use the one that fits your project.
How to use the new models
- Open the tool you need: the AI Song Generator, AI Music Generator, AI Video Generator or AI Music Video Generator.
- Pick your model from the model dropdown.
- Write your prompt, or use the same brief across several models to compare them.
- Generate, compare and keep the version you like best.
- Open it in AI Studio to edit, arrange and mix.
Or just ask Agent One
Agent One can pick the right model for the task, or use the one you name. "Make a short Halloween clip with Veo 3.1" works as well as clicking through the menus.
Try the Halloween brief yourself
Copy the briefs from the video, run them through the models you want, and compare. It's the quickest way to learn which model you'll reach for next time.
Who this is for
- Artists and producers comparing sounds before they commit to one.
- Video creators testing motion, realism and editing quality.
- Anyone making seasonal content who wants the right model for each job without juggling apps.
FAQ
Which new song models are on Soundverse?
Google Lyria 3.5 and ElevenMusic v2.5.
Which new video models are on Soundverse?
Veo 3.1, Seedance 2.5, Seedance 2.5 Edit and Gemini Omni Flash 1.1.
How do I choose a model?
Pick it from the model dropdown in the generator you're using, or ask Agent One to choose one or use the one you name.
Can I compare models on the same prompt?
Yes. Run the same brief through several models and keep the one you like best.
Can I edit the results?
Yes. Open them in AI Studio to edit, arrange and mix.
Try it
Open Soundverse, pick a model from the dropdown, and give it your own brief.
