Is Gemini 3.8 Flash TTS free? Free tier vs paid prices, Oct 2026

Partly. Google's pricing page lists Gemini 3.8 Flash TTS with a free tier on the standard tier, input and output both "free of charge", and a paid tier at $0.50 per million text tokens in and $9.00 per million audio tokens out through December 31, 2026. From January 1, 2027 the paid standard prices double to $1.00 and $18.00.

So a free key is enough to audition a script and a voice. A production run that you bill to a client needs the paid tier, and it needs a budget that survives the January change.

What Google lists, tier by tier

The model id is gemini-3.8-flash-tts, generally available since September 22, 2026 per the release notes. The pricing page gives four rows for the paid side. Output is the audio you hear, and Google states that audio tokens correspond to 25 tokens per second of audio.

What a prototype costs once the free tier ends

At 25 audio tokens per second, one minute of speech is 1,500 output tokens. At the 2026 standard rate of $9.00 per million that is $0.0135 for the audio, and $0.027 at the 2027 rate of $18.00. Ten minutes of narration is 15,000 tokens, or $0.135 now. Those figures are the audio side only. The text you send in is billed separately at the input rate.

The pricing page lists batch and flex as having no free tier. The page I read does not state request limits or how free-tier content may be used, so check both in Google AI Studio before you send a client's unreleased script through a free key.

The same script on a per-character meter

Sume's TTS 1.0 is metered per transcript character instead: $0.0475 per 1,000 characters, which is the provider list rate times 1.25, rounded up to the cent per job, with spaces and punctuation counted and a 20,000-character maximum per request. A 500-character line is $0.02375, so it bills 3 cents. A 2,000-character script bills 10 cents. There is no tier to switch on, and the rate does not change on January 1.

That makes the comparison a units question. Count your characters, estimate your seconds, and price each. If you need Google's 130-language coverage or its voice design, the Gemini route is the one that has them. If you need a flat per-job quote you can show a client before you run it, the character meter is easier to quote.

A sensible order of work

Use the free tier for the first drafts, then price the final script on both meters before committing.

• Audition the script and two or three voices on the free tier, and keep the winning voice name in your notes.
• Count the finished script's characters and estimate its seconds from a first take, not from a guess.
• Price it at the paid 2026 rate, the 2027 rate, and the batch rate if it can wait.
• If the job is a reel or a promo, add the video and captions to the total, since the voice is rarely the largest line.