Humanness Index™ · TTS model
by ElevenLabs
Turbo v2 was ElevenLabs' first generation low latency model, an English only engine built to bring conversational agents under a latency bar its earlier models could not meet.
Standings as of Jun 13, 2026, 00:15 UTC
A real arena clip: a cloned source voice reading a customer support prompt at phone quality.
Turbo v2 was ElevenLabs' first generation low latency model, an English only engine built to bring conversational agents under a latency bar its earlier models could not meet. ElevenLabs now recommends the Flash family for new builds and describes Turbo v2 as functionally equivalent to Flash v2 but slower on average. It remains widely deployed in voice stacks built before Flash arrived, and it still posts a strong humanness showing on the Index.
Sources: elevenlabs.io
Released in late 2023, Turbo v2 defined the first generation of the low latency tier ElevenLabs later rebuilt as Flash. In our 50 trial streaming benchmark it returned first audio in a median of 302 ms, the slowest of the four low latency ElevenLabs models on the Index.
Sources: elevenlabs.io
Standings as of Jun 13, 2026, 00:15 UTC
| Rank | Provider | Model | Humanness | Latency |
|---|---|---|---|---|
| #12 | MiniMax | Speech 2 Turbo | 63 | 315 ms |
| #13 | Neuphonic | neu_hq | 63 | 276 ms |
| #14 | ElevenLabs | Turbo v2 | 63 | 302 ms |
| #15 | Inworld | TTS-1.5-max | 61 | 337 ms |
| #16 | ElevenLabs | Multilingual v2 | 61 | 1006 ms |
The benchmark is open source. Suggest a model, read the methodology, or ask us to put your voice in the arena.
