Humanness Index™ · Provider
ElevenLabs is the incumbent voice platform, with model families that span ultra low latency conversational agents to maximum expressiveness.
Standings as of Jul 28, 2026, 01:54 UTC
| Rank | Model | Humanness | Latency | Languages | Price / 1M chars |
|---|---|---|---|---|---|
| #10 | Turbo v2 | 75 | 302 ms | English | $50 |
| #9 | Flash v2 | 76 | 226 ms | English | $50 |
| #14 | Flash v2.5 | 68 | 197 ms | 32 | $50 |
| #1 | Eleven v3 | 96 | 758 ms | 70+ | $100 |
| Rank | Model | Humanness | Latency | Languages | Price / 1M chars |
|---|---|---|---|---|---|
| — | Turbo v2.5 | — | 265 ms | 32 | $50 |
| — | Multilingual v2 | — | 1006 ms | 29 | $100 |
ElevenLabs is the incumbent voice platform, with model families that span ultra low latency conversational agents to maximum expressiveness. Latency and language coverage vary sharply across those families, so every ElevenLabs model on the Humanness Index™ carries its own figures rather than one provider-wide number.
Sources: elevenlabs.io
Four ElevenLabs models compete in the arena, spanning the Turbo, Flash, and v3 families; the Multilingual v2 and Turbo v2.5 entries are retired from battles but keep their pages. ElevenLabs bills in credits: Flash and Turbo models bill one credit per two characters, while Eleven v3 and Multilingual v2 bill one credit per character, which is why those two carry a higher effective rate.
Sources: elevenlabs.io, elevenlabs.io
Compare the models in blind tests, read the methodology, or get in touch.
Build a TTS model? Add yours to the Index.
