Humanness Index™ · TTS model
by Inworld
TTS-1.5-max is the flagship of Inworld's 1.5 generation, launched in late 2025 with P90 first chunk latency under 250 ms across 15 languages.
Standings as of Jul 28, 2026, 04:12 UTC
A real arena clip: a cloned source voice reading a customer support prompt at phone quality.
TTS-1.5-max is the flagship of Inworld's 1.5 generation, launched in late 2025 with P90 first chunk latency under 250 ms across 15 languages. Inworld measured the generation at 30 percent more expressive than its predecessor with a 40 percent lower word error rate, and the 1.5 line reached the top of the Artificial Analysis Speech Arena ahead of Google and ElevenLabs. It targets the quality and speed balance most production agents need.
Sources: docs.inworld.ai
The quality and speed flagship of the 1.5 generation, with a mini sibling for the lowest latency. In our 50 trial streaming benchmark it returned first audio in a median of 337 ms including network time.
Sources: docs.inworld.ai
Standings as of Jul 28, 2026, 04:12 UTC
| Rank | Provider | Model | Humanness | Latency |
|---|---|---|---|---|
| #6 | xAI | Grok TTS (Streaming) | 86 | 285 ms |
| #7 | Speechify | Simba 3.2 | 83 | — |
| #8 | Inworld | TTS-1.5-max | 78 | 337 ms |
| #9 | ElevenLabs | Flash v2 | 76 | 226 ms |
| #10 | ElevenLabs | Turbo v2 | 75 | 302 ms |
Compare the models in blind tests, read the methodology, or get in touch.
Build a TTS model? Add yours to the Index.
