Google Lyria 3 Pro Preview is a multimodal model supporting text and images as input, with text and audio output. Its million-token context window is useful for large creative briefs, campaign references, and extended source material, while audio generation creates possibilities beyond conventional copy workflows.

For an agency, the strongest fit is early-stage concept development, multimedia ideation, and projects where written prompts need to connect with audio output. Vision support may help interpret visual brand references. The preview designation and absence of published capability measurements make it difficult to validate reasoning, coding, instruction following, or dependable tool-driven work. Treat it as an experimental creative component rather than a primary client-facing language model until reliability and API workflow behaviour are established.

Assessed August 9, 2026

Rankings consider pricing, capabilities, benchmarks, and real-world applicability and are refreshed as new models launch. Feedback?