Gemini 3.8 Flash TTS costs $0.50 per million text input tokens and $9 per million audio output tokens, which works out to about $0.81 per hour of generated speech. The cheaper Gemini 3.8 Flash-Lite TTS charges $6 per million audio tokens, or about $0.54 an hour. Both are introductory prices that run through December 31, 2026; on January 1, 2027, Google’s price page shows every rate doubling, to about $1.62 and $1.08 an hour.
Google released both models on September 23, 2026, in the Gemini API and Google AI Studio, as stable replacements for the older Gemini 3.1 Flash TTS preview. Below is the full price list, how the per hour math works, what you get for the money, and how the cost compares with OpenAI, Mistral and ElevenLabs.
Quick answer
- Flash TTS: about $0.81 per hour of audio now, about $1.62 from January 1, 2027
- Flash-Lite TTS: about $0.54 per hour now, about $1.08 from January 1, 2027
- Batch and Flex: half price, so roughly $0.41 and $0.27 per hour until the end of 2026
- Text input is almost free: an hour of narration needs only around 12,000 text tokens, well under one cent
- Free tier: standard requests are free of charge in Google AI Studio, with rate limits
How Google turns tokens into cost per hour
Google bills generated speech in audio tokens. Its pricing page states that audio equals 25 tokens per second, so one hour of output is 90,000 tokens. At $9 per million, 90,000 tokens cost $0.81. At $6 per million, they cost $0.54. Google also lists the price per 10 seconds of audio: $0.00225 for Flash TTS and $0.0015 for Flash-Lite TTS at the standard tier.
The text you send is billed separately at $0.50 per million tokens, but it barely registers. A narrator speaking at 150 words per minute covers about 9,000 words in an hour, roughly 12,000 text tokens, which costs about half a cent. In practice the audio output rate is the whole bill.
Full Gemini 3.8 TTS price list
| Model and tier | Text input per 1M | Audio output per 1M | Approx. per hour | From Jan 1, 2027 per hour |
|---|---|---|---|---|
| Flash TTS, Standard | $0.50 | $9.00 | $0.81 | $1.62 |
| Flash TTS, Batch or Flex | $0.25 | $4.50 | $0.41 | $0.81 |
| Flash TTS, Priority | $0.90 | $16.20 | $1.46 | $2.92 |
| Flash-Lite TTS, Standard | $0.50 | $6.00 | $0.54 | $1.08 |
| Flash-Lite TTS, Batch or Flex | $0.25 | $3.00 | $0.27 | $0.54 |
| Flash-Lite TTS, Priority | $0.90 | $10.80 | $0.97 | $1.94 |
| 3.1 Flash TTS Preview, Standard (older) | $1.00 | $20.00 | $1.80 | No intro pricing |
Source: Google AI for Developers, Gemini API pricing page, checked November 3, 2026. Per hour figures use Google’s 25 audio tokens per second and exclude the small text input cost.
The table shows why the old preview model is no longer worth using: Flash TTS at standard pricing costs less than half of what 3.1 Flash TTS Preview charges, even after the January increase.
What you get: Flash TTS versus Flash-Lite TTS
Google split the release by job. Flash TTS is built for creative direction and character work, such as games, audiobooks and podcasts. Flash-Lite TTS trades some of that control for a lower price and targets high volume dubbing and voice agents. Both share the same API schema, so switching between them means changing one model name.
| Feature | Gemini 3.8 Flash TTS | Gemini 3.8 Flash-Lite TTS |
|---|---|---|
| Launch | September 23, 2026, stable | September 23, 2026, stable |
| Languages | 130, with automatic language detection | 101 listed, automatic detection |
| Voices | 30 original voices plus a library of 2,000+ | Same library and voice options |
| Voice design from a text prompt | Yes | Yes |
| Voice replication | Yes, with consent recording | Yes, with consent recording |
| Per request limits | 8,192 input tokens, 16,384 output tokens | 8,192 input tokens, 16,384 output tokens |
| Google products using it | Gemini Notebook | Google Vids |
Sources: Google launch post, Gemini API model pages, LiteLLM day zero notes and eesel AI’s review.
Voice design and voice replication
Voice design lets you describe a voice in plain language, for example an older narrator with a slow, deliberate pace, and save the result as a reusable voice ID. Voice replication copies a real voice from a 30 second sample. Google requires a verbal consent recording from the voice owner that matches the reference speaker, and replication through AI Studio is not offered in Illinois, Texas, the EEA, the UK, Switzerland or India. eesel AI’s review reports that stored voice IDs are capped at 200 per project with a one year lifetime.
Performance controls and safety
Google lists line by line direction, long form generation, native two speaker scenes and nonverbal cues such as laughs, sighs and gasps. Every clip from Gemini audio models carries an inaudible SynthID watermark, and replicated voices also get C2PA content credentials. That matters if you publish audio commercially, since platforms increasingly check for provenance labels.
How the price compares with other TTS APIs
Rivals bill in different units, mostly per character, so per hour figures are estimates. The comparison below uses tokencost.app’s assumption of 150 words per minute for character billed services.
| Service | Billing unit | Approx. cost per hour |
|---|---|---|
| Gemini 3.8 Flash-Lite TTS | $6 per 1M audio tokens | $0.54 (intro) |
| OpenAI TTS-1 | $15 per 1M characters | about $0.74 |
| Mistral Voxtral TTS | $16 per 1M characters | about $0.79 |
| Gemini 3.8 Flash TTS | $9 per 1M audio tokens | $0.81 (intro) |
| OpenAI gpt-4o-mini-tts | $12 per 1M audio tokens | about $0.90 |
| OpenAI TTS-1 HD | $30 per 1M characters | about $1.49 |
| ElevenLabs | Monthly plans, $22 for 121,000 characters on Creator | roughly $8 to $10 inside a plan |
Competitor figures from OpenAI’s model documentation and tokencost.app’s TTS price comparison; ElevenLabs estimate divides plan price by included audio. Verify current rates before purchase.
The takeaway: at introductory rates, Flash-Lite TTS is the cheapest major hosted option on this list, and Flash TTS sits in the same band as OpenAI’s basic voices. After January 1, Flash TTS at $1.62 an hour becomes more expensive than OpenAI TTS-1 and Voxtral, so the case for it rests on quality, language coverage and voice design rather than price. OpenAI’s developer documentation now marks gpt-4o-mini-tts as deprecated, which is worth checking if you planned to use it as your fallback.
Latest update, November 2026
As of November 3, Google’s pricing page still lists the introductory rates through December 31, 2026, with the doubled prices starting January 1, 2027. Google has not announced an extension. The same end date applies to the Gemini 3.6 and 3.8 Flash text models, so many teams face two price increases in the same billing cycle. Google’s launch post lists voice remixing (adjusting timbre, pitch, pace and accent) as coming soon, and Gemini Enterprise availability as coming soon for both models.
On quality, Google reports that Flash TTS ranks first on Hume AI’s Voice Design Benchmark with 71.4. eesel AI notes that Hume’s founder is credited on that announcement, and that on the blind Provider Voice Arena Flash TTS placed second, behind Cartesia’s Sonic 3. Treat the benchmark claims as vendor reported until more independent tests appear.
Practical advice before you build
Checklist
- Budget at the 2027 rate: about $1.62 per hour for Flash TTS and $1.08 for Flash-Lite TTS
- Use Batch or Flex for audiobooks, back catalog dubbing and anything that does not need an instant reply; it halves the cost
- Split long scripts: 16,384 output tokens is about 11 minutes of audio per request
- Pick Flash-Lite for volume and test whether listeners notice a difference before paying 50% more for Flash
- Plan real time voice separately: these models do not support the Live API
- Keep consent records for any replicated voice, and check the regional restrictions
Tip
Try both models free in Google AI Studio with your own script before writing code. If you are new to Google’s assistant, start with our guide on how to use Google Gemini.
Synthetic voices are also spreading into consumer products. Google’s own audio summaries are one example, compared in our test of ChatGPT Study Mode versus NotebookLM, and Apple is rebuilding Siri around large models, as explained in our piece on the iOS 27 Siri AI waitlist. If the terminology is new to you, our plain English guide to what AI is covers the basics. More model coverage lives in our AI and software section.
FAQ
How much does Gemini 3.8 Flash TTS cost per hour?
About $0.81 per hour of generated audio at the standard tier through December 31, 2026, based on $9 per million audio tokens and 25 tokens per second. From January 1, 2027, it is about $1.62 per hour.
What is the difference between Gemini 3.8 Flash TTS and Flash-Lite TTS?
Flash TTS is built for creative direction in games, audiobooks and podcasts and costs $9 per million audio tokens. Flash-Lite TTS targets high volume dubbing and voice agents at $6 per million, about $0.54 per hour. Both use the same API schema.
When does Gemini TTS pricing go up?
Google’s pricing page shows the introductory rates ending December 31, 2026. Every Gemini 3.8 TTS rate doubles on January 1, 2027.
Is Gemini 3.8 Flash TTS free to try?
Yes. Standard requests are free of charge on the free tier in Google AI Studio, with rate limits. Batch processing is not available on the free tier.
Can Gemini 3.8 Flash TTS clone a voice?
Yes. Voice replication works from a 30 second sample, but Google requires a verbal consent recording from the voice owner, and replication in AI Studio is not available in Illinois, Texas, the EEA, the UK, Switzerland or India.
How many languages does Gemini 3.8 Flash TTS support?
Flash TTS supports 130 languages with automatic language detection. Flash-Lite TTS lists 101 languages.
Sources
- Google AI for Developers: Gemini API pricing
- Google: Gemini 3.8 text to speech launch post
- Google AI for Developers: Gemini 3.8 Flash-Lite TTS model page
- Google AI for Developers: Speech generation guide
- LiteLLM: Day zero support for Gemini 3.8 Flash TTS and Flash-Lite TTS
- eesel AI: Gemini 3.8 Flash TTS review
- OpenAI: GPT-4o mini TTS model documentation
- tokencost.app: TTS API pricing compared
About this article: GeekBlog covers U.S. technology news, AI, phones, smartwatches and gaming. Every story is written and checked under our Editorial Policy. Spotted a mistake or have a story tip? Contact our editors.
