Rather than outputting speech, this model generates dynamic instrumentation, arrangement, and vocals across any genre from a simple text prompt.

The cheapest and fastest Text-to-Speech (TTS) model available. It cuts API character credit costs by 50% but sacrifices a minor amount of emotional depth compared to Turbo or standard models.

Optimized for low latency while maintaining high voice generation quality. It provides the exact same language coverage as Flash v2.5 but with cleaner vocal textures

A legacy powerhouse for heavy content creation. It handles long-form paragraphs with the highest stability against performance drift, though it lacks newer languages added to v2.5 like Hungarian, Norwegian, and Vietnamese