Google releases Lyria 3 Pro Preview for full-length music generation
Google has released Lyria 3 Pro Preview, a music generation model capable of producing full-length songs with verses, choruses, bridges, vocals, and timed lyrics from text prompts or images. The model features a 1,048,576 token context window and charges $0.08 per generated song through the Gemini API.
Google Lyria 3 Pro Preview — Quick Specs
Google Releases Lyria 3 Pro Preview for Full-Length Music Generation
Google has released Lyria 3 Pro Preview, a music generation model that produces complete songs with structural coherence, including vocals, timed lyrics, and full instrumental arrangements.
Key Specifications
Lyria 3 Pro Preview launched on March 30, 2026. The model accepts text prompts or images as input and generates high-quality 48kHz stereo audio output. According to Google, the model can generate full-length songs featuring verses, choruses, and bridges with maintained musical structure.
The model supports a context window of 1,048,576 tokens. Pricing is structured differently from traditional token-based models: $0.08 per generated full-length song. Input and output token pricing shows as $0/M, reflecting Google's song-based pricing model rather than token-based billing.
Distribution and Access
Lyria 3 Pro Preview is available through the Gemini API. The model is also accessible via OpenRouter, which provides unified API access across multiple providers and handles request routing for optimal uptime and performance.
OpenRouter's integration normalizes requests and responses, allowing developers to call Lyria 3 Pro Preview using OpenAI-compatible SDK calls or directly through the OpenRouter API. The platform tracks usage metrics and provides uptime statistics across providers.
Capabilities
The model handles both text-to-audio and image-to-audio generation workflows. Google claims Lyria 3 Pro delivers "structural coherence" in generated songs, a technical specification addressing a known limitation in earlier music generation systems where songs could lose harmonic and rhythmic consistency across sections.
The inclusion of "Pro" in the name suggests a higher-capability tier within the Lyria 3 family, implying standard and potentially other variants may exist with different feature sets or pricing.
What This Means
Google's release of Lyria 3 Pro Preview marks a shift toward production-ready music generation. The fixed $0.08-per-song pricing eliminates variable costs based on song length, potentially making long-form music generation more economical than token-based pricing would allow. However, the "Preview" designation indicates this remains a limited availability release, suggesting broader availability may require additional refinement or capacity planning by Google. For music applications, content creators, and developers building audio tools, this provides a new option from a major cloud provider with significant infrastructure backing.
Related Articles
NVIDIA Releases Nemotron 3.5 Lightning: 30B MoE Model with 1M Token Context and 3B Active Parameters
NVIDIA released the full-precision BF16 reference weights for Nemotron 3.5 Lightning, a 30B-parameter Mixture-of-Experts model with only 3B active parameters and support for up to 1 million tokens of context. The model uses a hybrid Mamba-2, MoE, and Attention architecture and is licensed under OpenMDW-1.1 for commercial use.
Qwen Releases Qwen3.8 2.4T A95B, a 2.4-Trillion-Parameter Open-Weight MoE Model
Qwen has released Qwen3.8 2.4T A95B, an open-weight sparse mixture-of-experts model with 2.4 trillion total parameters and 95 billion active parameters per forward pass. The model is the open-weight variant of Qwen3.8 Max, targeting coding, research, complex reasoning, and agentic workflows with a 262K token context window.
xAI Releases Grok 4.6, a 1.5T-Parameter Model Powering New 'Grok Bot' AI Teammate Product
xAI released Grok 4.6, a confirmed 1.5T-parameter model built on Grok 4.5 with heavier training on long-horizon agentic tasks. It powers the newly launched Grok Bot product and scores 61 on Artificial Analysis's Intelligence Index at $2/$6 per 1M input/output tokens — well below frontier competitors.
Alibaba Releases Qwen3.8-2.4T-A95B-FP8: 2.4T-Parameter Open Model with 1M-Token Context
Alibaba's Qwen team has released Qwen3.8-2.4T-A95B-FP8, an open-weight, FP8-quantized MoE model with 2.4 trillion total parameters and 95 billion activated per token. It natively supports 262,144 tokens of context, extensible to 1,010,000, and forms the base for the hosted Qwen3.8-Max API.
Comments
Loading...