Gemini 3.8 TTS Is the Cheapest Studio-Grade Voice Right Now — and It Doubles on 1 January

We have no referral relationship with Google, so nothing on this page is sponsored. Links to Murf and ElevenLabs are referral links. Every rate below was read off each vendor’s own pricing page on 27 September 2026.

Google made its new text-to-speech models generally available on 22 September, and on price they undercut almost everything else on the market. There is a catch, and Google states it plainly in the pricing table rather than hiding it: the rates you see now are promotional. On 1 January 2027 every Gemini TTS price doubles.

The published rates, now and from January

Flash TTS is the one Google describes as engineered for studio-grade fidelity, expressive delivery and long-form stability — audiobooks, narration, character work. Flash-Lite TTS is tuned for high throughput and low latency, which is what you want behind a voice agent or an app feature. Both ship with voice design, voice replication with consent verification, and more than 150 prebuilt and custom voices.

What that costs per hour of audio

Token prices are hard to compare with the per-minute pricing the rest of the voice market uses, so here is the conversion. Google’s own pricing page states that audio tokens correspond to 25 tokens per second — 1,500 tokens a minute, 90,000 an hour. Multiply that out:

The Murf figures come from our Murf pricing breakdown and assume roughly 900 characters of script per minute of speech. Google’s figures are arithmetic from its published token rates and its own 25-tokens-per-second note, not a quoted per-minute price.

Two things stand out. Today, Gemini Flash-Lite TTS is priced level with Murf’s cheapest API voice, and Flash TTS — the expressive one — costs half what Murf charges for its comparable Gen2 model. In January those positions reverse: Gemini Flash TTS lands exactly where Murf Gen2 sits now, and Flash-Lite doubles to twice Falcon’s rate.

Against subscription tools the gap is much wider. A Murf Studio Creator plan works out around $9.50 an hour of finished audio; even after January’s increase, Gemini’s studio model is roughly a sixth of that. You are trading an editor, timeline syncing and stock media for raw API access — the same trade we described in the Murf pricing piece.

How it compares with ElevenLabs

ElevenLabs prices consumer plans by credits rather than tokens, and it did not change prices in September. Its Starter plan is $6 a month and Creator $22, with the practical limits — credit rollover, professional voice cloning gated to Creator, dubbing minutes counted separately — laid out in our pricing tracker.

The honest comparison is not price but product. ElevenLabs sells a finished platform: a studio interface, dubbing, agents, a voice library with licensing attached. Google sells an endpoint. If you need a voice inside software you are already building, Gemini TTS is now the cheapest serious option. If you need to produce audio without writing code, it is not really competing for your money at all. Both sit in our tool picker.

What the deadline means in practice

• Model your costs at the January rate, not today’s. A pipeline that is comfortably profitable at $0.54 an hour and marginal at $1.08 is a pipeline with a problem in the new year, not a bargain now.
• Batch what you can. Anything that does not need to be generated in real time — audiobook chapters, course narration, bulk voiceover — is half price through the batch endpoint, and stays half price relative to standard rates after January.
• Do not rebuild on price alone. A promotional rate with a published end date is a weak reason to migrate an existing pipeline. It is a good reason to try the models now while the trial is cheap.
• Watch for the same pattern elsewhere. Google is running the same promotional structure on Gemini 3.8 Flash text ($0.75/$3.75 today, $1.50/$7.50 from January). Dated introductory pricing across a model family is worth treating as the normal price arriving late.

The verdict

Credit where it is due: Google published the increase in the same table as the current price, with a date. That is more honest than the silent repricing this industry usually does, and it lets you plan. Use the next three months to test whether Gemini TTS is good enough for your use case — but build your budget on the 2027 numbers, because those are the real ones.

If you would rather buy finished audio than build a pipeline, compare the alternatives on Murf’s pricing page or with a free ElevenLabs account, and check our tracker for what each one actually costs per minute.

All Gemini rates verified on Google’s Gemini API pricing page on 27 September 2026, including the stated 25 audio tokens per second and the 1 January 2027 changes. Murf and ElevenLabs figures from their own pricing pages, last verified 21 September 2026.