Data read October 27, 2026 · Coval, Speko, Artificial Analysis, Cekura
TTS API pricing in 2026: list price per 1 million characters
Among ranked models, list prices run from $6.60 (SpeechifyAI Simba 3.0) to $57.80 (Gradium TTS) per million characters, as printed by the benchmarks on October 27, 2026.
Prices are not read from vendor websites: each comes from one of the four benchmarks, and a range appears when two benchmarks print different figures. 3 ranked models have no price in any source; they are listed last.
How the rows are ordered. Ordered by lowest list price, cheapest first; models without a price come last, in ranking order.
| Model | Price per 1M characters List price; a range when sources differ | Rank | Speed tier From the Coval P95 | P95 (ms) 1 turn in 20 is slower | Listener Elo Artificial Analysis, with 95% interval |
|---|---|---|---|---|---|
 SpeechifyAI Simba 3.0 | $6.60 to $10 | 5 | Fast enough | 149.4 | 1,116 ±11 |
 Alibaba Qwen3 TTS Fast | $10 | 1 = | Headroom | 98.8 | n/a |
 Murf AI Falcon 2 | $10 | 12 = | Visible delay | 363.5 | 1,154 ±12 |
 Inworld Realtime TTS-2 Flash | $10.40 to $15 | 1 = | Headroom | 100.9 | 1,218 ±13 |
 Palabra TTS v1 | $12 | 7 | Fast enough | 214.3 | n/a |
 Soniox TTS RT v1 | $13 to $14.10 | 10 = | Visible delay | 319.0 | n/a |
 Soniox TTS Real-Time v2 | $13 to $14.81 | 10 = | Visible delay | 324.7 | 1,180 ±11 |
 Alibaba Qwen3 TTS Flash Realtime | $13 | 21 | Visible delay | 785.8 | n/a |
 Inworld Realtime TTS-2 | $15 to $25 | 6 | Fast enough | 194.7 | 1,267 ±14 |
 SpaceXAI Grok TTS | $15 | 18 | Visible delay | 473.9 | 1,155 ±13 |
 Fish Audio S2.1 Pro | $15 | 19 | Visible delay | 533.0 | 1,184 ±13 |
 Smallest.ai Lightning V3.1 Pro (Jul 2026) | $19.50 | 20 | Visible delay | 703.3 | 1,169 ±12 |
 Rime Mist v3 | $30 | 12 = | Visible delay | 359.9 | n/a |
 Google Chirp 3: HD | $30 | 22 | Visible delay | 917.2 | 1,061 ±12 |
 Gradium TTS | $36 to $57.80 | 1 = | Headroom | 104.3 | 1,174 ±13 |
 Cartesia Sonic 3.6 | $39.20 to $50 | 14 | Visible delay | 374.4 | 1,292 ±14 |
 Cartesia Sonic 3.5 | $39.20 to $50 | 15 | Visible delay | 390.5 | 1,190 ±11 |
 ElevenLabs Eleven v4 Turbo | $40 | 8 | Fast enough | 238.6 | 1,333 ±16 |
 ElevenLabs Flash v2.5 | $40 to $50 | 9 | Fast enough | 255.1 | 1,076 ±11 |
 Rime Coda | $50 | 16 = | Visible delay | 428.1 | 1,054 ±12 |
 Alibaba Qwen3 TTS 1.7b | n/a | 4 | Fast enough | 136.6 | n/a |
 Deepdub PhantomZ 4.0 | n/a | 16 = | Visible delay | 430.0 | n/a |
 Fish Audio S2.1 Pro Free | n/a | 23 | Visible delay | 1,137.9 | n/a |
Frequently asked questions
- How much does a multilingual TTS API cost?
- For ranked models, list prices run from $6.60 to $57.80 per million characters. The cheapest is not the fastest: compare the price column with the tier and P95 beside it.
- Which TTS API is cheapest per million characters?
- Starting with the cheapest, SpeechifyAI Simba 3.0, Alibaba Qwen3 TTS Fast, and Murf AI Falcon 2. The order uses the lowest price any benchmark prints for the model.
- Why do some models show a price range?
- Coval, Speko, Artificial Analysis and Cekura do not always print the same price. The range shows the lowest and highest figures; the model page lists each source.
- Do these prices include volume discounts?
- No. They are list prices as benchmarks print them. Enterprise or committed-use discounts are not covered.