Google launches two benchmark-topping speech generation models
Google's Gemini 3.8 Flash TTS models top every speech-generation benchmark—and cost half as much as competitors.
Why it matters
Two new text-to-speech models from Google set capability benchmarks while offering a cost/quality tradeoff that practitioners will evaluate for production voice workloads. The dual-model release (performance vs. efficiency) signals how frontier labs are now shipping capability ladders rather than single releases.
The key facts
12 to knowTwo new models: Gemini 3.8 Flash TTS (best quality) and Gemini 3.8 Flash-Lite TTS (cost/speed optimized)
Both models benchmark-topping in speech generation
Unified API design for side-by-side developer use
Available on Google Cloud Platform
Flash-Lite TTS optimized for inference speed and cost efficiency
Released September 23, 2026
Two new models: Gemini 3.8 Flash TTS (quality-optimized) and Gemini 3.8 Flash-Lite TTS (cost/speed-optimized)
Both models report benchmark-topping performance
Unified API across both models simplifies side-by-side deployment
Available through Google Cloud platform
Flash-Lite TTS targets inference speed and cost efficiency
Flash TTS prioritizes audio quality
Go to the source
SiliconAnglesiliconangle.com
Publisher excerpt: Google LLC today made two new text-to-speech models, Gemini 3.8 Flash TTS and Gemini 3.8 Flash-Lite TTS, available through its cloud platform. The algorithms have highly similar application programming interfaces, which makes using them side-by-side relatively simple for developers. Flash-Lite TTS…