Google launches Gemini 3.8 Text-to-Speech API with 24 voices

Google has added Gemini 3.8 Text-to-Speech to its Cloud AI Platform APIs, with 24 neural voice profiles. Pricing starts at $0.000016 per character, which works out to about $16 per million characters. That is a competitive price for audio generation at scale.

The model targets developers building voice assistants, narration and accessibility features. Those are also the products Salesforce is paying a reported $2.7 billion to get through Cresta, and that xAI is pursuing with its new Voice APIs. Google's pitch is integration with the rest of the Gemini stack plus aggressive per-character pricing.

The main competitors are OpenAI's TTS models, ElevenLabs, Amazon Polly and Azure AI Speech. The market increasingly competes on naturalness, latency and price per character rather than raw voice count.

The caveat is that the details come mainly from a third-party write-up. Developers should check Google's official documentation for regional availability, latency and language coverage before migrating production workloads.