Text-to-speech API pricing compared
Pay-as-you-go text-to-speech APIs range from $0.004 to $0.144 per minute of audio, a 40× spread. The middle of the market sits around $0.027 a minute.
Pay-as-you-go prices
| API and voice tier | Per 1M characters | Per minute | Free each month |
|---|---|---|---|
| Google Cloud Standard voices Google lists these under legacy models. | $4 | $0.004 | 4,000,000 chars |
| Google Cloud WaveNet voices Listed at the same price and billing code as Standard. | $4 | $0.004 | 4,000,000 chars |
| Amazon Polly Standard First 12 months only: 5 million characters a month free, not counted here. | $4 | $0.004 | None |
| Murf API, Falcon | $10 | $0.009 | None |
| OpenAI tts-1 | $15 | $0.013 | None |
| Azure AI Speech Neural East US price; varies by region. Separate free tier (F0): 0.5 million characters a month, a different resource tier from pay as you go, so not counted here. | $15 | $0.013 | None |
| Deepgram Aura-1 | $15 | $0.013 | None |
| Fish Audio API, S2.1-Pro Billed per UTF-8 byte of text: one a character in plain English, more for accented and non-Latin letters. S2-Pro and S1 cost the same. | $15 UTF-8 bytes | $0.013 | With limits |
| Google Cloud Neural2 voices | $16 | $0.014 | 1,000,000 chars |
| Amazon Polly Neural First 12 months only: 1 million characters a month free, not counted here. | $16 | $0.014 | None |
| Gemini 2.5 Flash TTS A preview model on the Gemini API that Google says may be retired from 17 November 2026. | Billed in tokens | $0.015 | With limits |
| Gemini 3.8 Flash-Lite TTS About $0.009 a minute until 31 December 2026, an introductory price. | Billed in tokens | $0.018 | With limits |
| Azure AI Speech Neural HD East US price; varies by region. | $22 | $0.02 | None |
| Murf API, Gen2 | $30 | $0.027 | None |
| OpenAI tts-1-hd | $30 | $0.027 | None |
| Google Cloud Chirp 3: HD voices | $30 | $0.027 | 1,000,000 chars |
| Amazon Polly Generative First 12 months only: 100,000 characters a month free, not counted here. | $30 | $0.027 | None |
| Deepgram Aura-2 | $30 | $0.027 | None |
| Gemini 3.8 Flash TTS About $0.014 a minute until 31 December 2026, an introductory price. | Billed in tokens | $0.027 | With limits |
| Gemini 3.1 Flash TTS A preview model on the Gemini API that Google says may be retired from 17 November 2026. | Billed in tokens | $0.03 | With limits |
| Gemini 2.5 Pro TTS A preview model on the Gemini API that Google says may be retired from 17 November 2026. No free tier. | Billed in tokens | $0.03 | None |
| ElevenLabs API, Flash / Turbo | $40 | $0.036 | 10,000 chars |
| Resemble AI Flex pay-as-you-go $0.00067 per second of audio, so text length doesn't matter. | Billed per second | $0.04 | None |
| ElevenLabs API, v2 Multilingual | $80 | $0.072 | 10,000 chars |
| Amazon Polly Long-Form First 12 months only: 500,000 characters a month free, not counted here. | $100 | $0.09 | None |
| Google Cloud Studio voices | $160 | $0.144 | 1,000,000 chars |
Per-minute prices assume 900 characters of text per minute of speech for every vendor. Price only: quality, latency, languages and licences differ.
Free allowances change the picture for small volumes. Google's free tier covers about 4,444 minutes a month of Standard or WaveNet voices and 1,111 minutes of its newer voices, so a low-volume project may pay nothing there. Amazon Polly's free tier for Neural, Long-Form and Generative voices lasts only your first 12 months, so it isn't counted here.
Subscription API plans
Cartesia sells monthly credit bundles instead of pure pay-as-you-go. Used in full, they work out to:
| Plan | Per month | Characters included | Per minute if fully used | Extra, per 1M characters |
|---|---|---|---|---|
| Cartesia Pro | $5 | 100,000 | $0.045 | $65 |
| Cartesia Startup | $49 | 1,250,000 | $0.035 | $45 |
| Cartesia Scale | $299 | 8,000,000 | $0.034 | $38 |
Token-priced models
Some newer models bill in tokens rather than characters. Gemini's are in the table above, converted with the vendor's own audio tokens per second (how: Gemini TTS pricing). This one isn't, because its vendor states no conversion:
- OpenAI gpt-4o-mini-tts
- $0.60 per 1M text input tokens and $12.00 per 1M audio output tokens, at most 2,000 input tokens a request. OpenAI states no per-minute figure and no audio tokens per second for this model, so we can’t convert it to minutes honestly.
How many characters make a minute?
Vendors disagree, which is why we use one pace for everyone. Their own statements:
- ElevenLabs
- "$0.08 ~$0.08/minute 10,000 ~10 min"
- Murf
- "100K Characters ~ 1h 40m"
- Cartesia
- "1 credit per character, about 750 to 800 credits per minute of audio."
- Amazon Polly
- "1 million characters ~23 hours, 8 min"
At ElevenLabs' and Murf's pace a minute needs about 1,000 characters; at Amazon's, about 720. A vendor using the higher figure looks about 28% cheaper per minute for the same text. You can set your own pace in the text-to-speech cost calculator, or convert any length on how many characters is a minute of speech.
- AI audiobook cost calculatorNarrating a whole book: it prices it on every API and plan
- AI voice agent cost per minuteGiving a phone or voice agent a voice: the whole conversation, voice included
Unavailable vendors
- Play.ht
- On 5 October 2026 play.ht and play.ai did not load: the domains no longer resolve, and their documentation site was gone.
- LOVO (Genny)
- On 5 October 2026 every LOVO site returned a "deployment disabled" error, and its API docs were marked inactive.
- Which text-to-speech APIs are free, and for how longLooking for a free tier
Sources
14 pricing pages, read on 5 October 2026 unless noted
- Murf pricing
- Speechify Studio pricing
- WellSaid pricing
- Cartesia pricing
- OpenAI pricing
- Google Cloud pricing
- Azure AI Speech pricing
- Amazon Polly pricing
- Deepgram pricing
- Resemble AI pricing
- Gemini pricing, read on 7 October 2026
- Fish Audio pricing, read on 7 October 2026
- docs.fish.audio/developer-guide/models-pricing/pricing-and-rate-limits
- ElevenLabs API pricing
Prices on this page were checked against the official pricing pages on .