Text-to-speech API pricing, side by side

Every character-billed price we could verify, sorted cheapest first, with the official source for each. Prices read 2026-10-08; they change.

List price per million characters

Provider and modelUSD per 1M charactersSource
Speakvora catalog voices$0.70this site
Amazon Polly Standard$4.00source
Google Cloud Text-to-Speech Standard$4.00source
Speakvora your own voice$6.00this site
Deepgram Aura-1$15.00source
OpenAI tts-1$15.00source
Amazon Polly Neural$16.00source
Google Cloud Text-to-Speech Neural2$16.00source
Deepgram Aura-2$30.00source
Amazon Polly Generative$30.00source
Google Cloud Text-to-Speech Chirp 3 HD$30.00source
OpenAI tts-1-hd$30.00source
Cartesia Sonic, Scale overage$38.00source
ElevenLabs Flash / Turbo$40.00source
Cartesia Sonic, Pro plan credits$50.00source
ElevenLabs v2 Multilingual$80.00source
ElevenLabs v3$80.00source
Amazon Polly Long-Form$100source
Google Cloud Text-to-Speech Studio$160source

Not in the table because their native unit is not characters: OpenAI gpt-4o-mini-tts (tokens), Cartesia (credits; $37 to $65 per million characters effective across its plans and overage), and the ElevenLabs v4 promotion (a temporary discount, list $80).

How to compute your own bill

Monthly cost = characters per month ÷ 1,000,000 × price per million. A notification of 45 characters sent 100,000 times is 4.5 million characters. That is $3.15 at $0.70, $72 at $16 and $360 at $80. Add what the provider charges for anything else you use.

What the free allowances really are

ProviderFree allowance read on its pageCatch
Speakvora100,000 characters per accountOnce; not monthly
Amazon Polly5M Standard characters a month; 1M Neural for 12 monthsStandard is the older tier
Deepgram$200 creditA credit, not an allowance
CartesiaFree plan 20K creditsMonthly plan credits
ElevenLabsA Free plan existsAllowance not labelled on the page we read

Which one should you pick?

  • Inside Polly's free tier: use Polly Standard.
  • Mostly repeated English prompts: download a Speakvora pack, no per-play cost.
  • Premium expressive voices: ElevenLabs or OpenAI gpt-4o-mini-tts.
  • Low-latency voice agents on the phone: Cartesia or Deepgram.
  • Already on Google Cloud or AWS with committed spend: stay put until the commitment ends.

One-on-one pages: , , , , , .