Google's Gemini 3.8 TTS Pricing Doubles on Jan. 1, 2027

Google rolled out two new text-to-speech models, Gemini 3.8 Flash TTS and Gemini 3.8 Flash-Lite TTS, on Sept. 23, 2026, in the Gemini API and Google AI Studio. Flash TTS is built for scripted, creative work such as audiobooks and games; Flash-Lite TTS targets high-volume, real-time uses such as voice agents and replaces the earlier gemini-3.1-flash-tts-preview model.

Google's own pricing page lists standard-tier rates of $0.50 per million input tokens and $9 per million audio output tokens for Flash TTS through Dec. 31, 2026, then states both figures double to $1 and $18 on Jan. 1, 2027. Flash-Lite TTS follows the same input rate but a lower $6 output rate through the end of the year, also doubling in January. Batch-tier customers get roughly half the standard rate on each model, before and after the increase.

What actually changed from the last generation

The headline feature is scale: Google's preset voice library grew from about 30 voices to more than 2,000, spanning over 100 languages and dialects, according to Unite.AI's coverage of the rollout. Flash TTS can generate an original voice from a text description of role, accent and vocal character, and can clone an existing voice from a sample, though Google requires a spoken consent recording that the system checks against the reference speaker before the clone is created.

Voice replication through AI Studio is unavailable in Illinois, Texas, the European Economic Area, the United Kingdom, Switzerland and India, Unite.AI reported, a restriction tied to those jurisdictions' voice-likeness or biometric laws. Every clip the models generate carries Google's SynthID watermark, according to TheNextWeb's report on the launch.

Where Google says it ranks

Google's launch materials cite the Hume AI Voice Design Benchmark, an independent third-party evaluation, giving Flash TTS a 71.4 overall score and a 60.8 accent-modeling score, both the top result among models the benchmark tracks. Google also points to Hume's separate Overall Quality Index, where it says Flash TTS and Flash-Lite TTS placed first and second.

1. Flash TTS (through Dec. 31, 2026)9 $ per million audio output tokens
2. Flash TTS (from Jan. 1, 2027)18 $ per million audio output tokens
3. Flash-Lite TTS (through Dec. 31, 2026)6 $ per million audio output tokens
4. Flash-Lite TTS (from Jan. 1, 2027)12 $ per million audio output tokens

Source: Google, Gemini API pricing page, accessed 2026-09-25

Who is already building on it

Google named ten launch partners integrating the new models, including real-time voice infrastructure providers Agora, LiveKit and Pipecat, along with Figma, HeyGen and Vercel, per TheNextWeb's report. TheNextWeb also named ElevenLabs, India's Murf AI and Adobe Firefly's August-added speech tools as the competing products Gemini 3.8 TTS is priced against.

Leland Rechis, Google's group product manager for Gemini audio, and Alan Cowen, the company's director of research science, are named in the rollout materials Unite.AI cited, though neither is quoted directly on the January pricing change. The company has not said whether the January price increase applies retroactively to usage locked in under existing enterprise contracts, and its rollout notes list Gemini Enterprise access as "coming soon" without a date. Anyone pricing Flash TTS into a product now is, per Google's own page, budgeting for a rate that stops applying in about three months, a detail the rollout coverage largely left out in favor of the voice-count and benchmark figures. The Gemini API joins a wider push into agent tooling and infrastructure, including Google's orbital TPU testing, that the company has been rolling out through September.