Lyria 3 Clip Preview
Neural Network
Lyria 3 Clip Preview is a Google model for music generation from text descriptions, creating 30-second clips.
Specifications
- Input
- text, images, documents
- Output
- text
- Context size
- 1 048 576 tokens
- Max answer length
- 65 536 tokens
- Capabilities
- Batch API
How it works Lyria 3 Clip Preview?
Lyria 3 is a family of music generation models from Google, available via the Gemini API; the Clip Preview variant creates 30-second compositions from text descriptions. You define the mood, genre, tempo, and instrumentation in words, and the model returns a finished audio fragment, bypassing studios and long iterations. Through BotHub, the model works from Russia without a VPN or foreign card: payment is made with Russian cards in rubles, per actual generated clip, with no subscription. In the same window, you have access to over 250 models, so you can assemble a soundtrack, script, and cover in one go. For products, there is a unified API, traffic is encrypted, and companies have access to contracts, invoices, electronic document management, and an admin panel with team limits. It is useful if you are editing videos for social media and need a background track for the timing; making jingles and sound intros for podcasts and ads; assembling music for game scenes and prototypes; or looking for rough mood references before working with a composer.Frequently asked questions about Lyria 3 Clip Preview
You can use generated results for commercial purposes. You own all rights to the content you create. The only restriction: make sure your prompt does not include copyrighted third-party material. You are responsible for respecting the rights to any input data.
According to our data, this Google model accepts text, documents, and images as input and responds with text. It is suitable for analyzing files and screenshots, summarizing, answering questions based on content, rewriting, and structuring materials. Batch processing is available for large volumes of similar requests.
The output limit is 65,536 tokens, which is also the size of the context window. This is enough for a long document analysis, a detailed instruction, or a voluminous draft in its entirety, without splitting the task into parts. Note that input data and the response share the same context.
A separate reasoning mode is not noted in our data, so we will not promise it. In practice, a simple trick helps: ask in the prompt to break down the task step-by-step and show the thought process before the final conclusion — the quality of answers to complex tasks usually increases.
There is no mention of function calling or structured output in our data. If you need strict JSON, define the schema directly in the prompt and validate the response on your side. BotHub has a unified OpenAI-compatible API, so you can switch to a model with functions without rewriting the integration.
The model accepts images and documents — you can send a screenshot, diagram, scan, or file and ask to analyze the content. Audio and video input is not noted in our data. The result is returned as text in any case; generation of other formats is not claimed.
We have no data on languages, so we cannot make exact promises — test it on your real task. In BotHub, this is convenient: access from Russia without a VPN or foreign card, pay-as-you-go for tokens, and over 250 models for comparison in one window.