Google announced Gemini 3.1 Flash TTS on 15 April 2026, positioning the model as "the next generation of expressive AI speech" [source]. The release marks a significant architectural shift in Google's text-to-speech offerings, replacing the previous Gemini 2.0 Flash TTS model that launched in December 2024.

According to the blog post, Gemini 3.1 Flash TTS introduces enhanced prosody control, multi-speaker synthesis, and reduced latency compared to its predecessor. Google states the model supports 100+ languages and can generate speech with "natural emotional range" across contexts including customer service, audiobook narration, and accessibility applications.

The announcement does not provide comparative benchmarks against Gemini 2.0 Flash TTS or detail specific performance metrics. Google indicates the model is available through the Gemini API and Vertex AI, with pricing information referenced in separate documentation.

This release follows a pattern observed across major AI providers in early 2026, with OpenAI, Anthropic, and Microsoft each launching updated speech synthesis models in the first quarter. Google's decision to increment the version number from 2.0 to 3.1 suggests either a major capability expansion or a realignment of internal versioning schemes.

Users of the previous Gemini 2.0 Flash TTS model should verify compatibility and output characteristics, as model transitions historically introduce changes in voice timbre, pacing, and pronunciation handling. Google has not specified a deprecation timeline for the 2.0 variant.

The blog post includes audio samples demonstrating the model's output across multiple languages and speaking styles, though independent verification of claimed improvements remains pending.