Best text-to-speech models
Voice models ranked by head-to-head human preference.
Sonic 3.6 leads from Cartesia, ahead of 99 others.
- 1Cartesia
Sonic 3.6
1,285 elo
- 2SpeechifyAI
Simba 3.2
1,241 elo
- 3Alibaba
Qwen-Audio-3.0-TTS-Plus
1,240 elo
ranked by elo · 100 models · updated Aug 30 · 1–50
| # | Model | Creator | Elo |
|---|---|---|---|
| 1 | Sonic 3.6 | Cartesia | 1,285 |
| 2 | Simba 3.2 | SpeechifyAI | 1,241 |
| 3 | Qwen-Audio-3.0-TTS-Plus | Alibaba | 1,240 |
| 4 | Luna TTS | VUI Labs | 1,225 |
| 5 | Breeze TTS 2 | BreezeBlue | 1,218 |
| 6 | v3 Conversational | ElevenLabs | 1,215 |
| 7 | Gemini 3.1 Flash TTS | 1,209 | |
| 8 | StepAudio 2.5 TTS | StepFun | 1,208 |
| 9 | Sonic 3.5 | Cartesia | 1,198 |
| 10 | Realtime TTS 1.5 Max | Inworld | 1,198 |
| 11 | Lightning V3.1 Pro (Jul 2026) | Smallest.ai | 1,196 |
| 12 | Realtime TTS-2 - Research Preview | Inworld | 1,186 |
| 13 | Soniox TTS Real-Time v2 | Soniox | 1,181 |
| 14 | Eleven v3 | ElevenLabs | 1,179 |
| 15 | Speech 2.8 HD | MiniMax | 1,172 |
| 16 | Falcon 2 | Murf AI | 1,158 |
| 17 | Gradium TTS Beta | Gradium | 1,154 |
| 18 | Speech 2.8 Turbo | MiniMax | 1,152 |
| 19 | Async Flash v1.5 | async | 1,146 |
| 20 | Fish Audio S2.1 Pro | Fish Audio | 1,146 |
| 21 | Step TTS 2 (Mar 2026) | StepFun | 1,140 |
| 22 | Speech 2.6 HD | MiniMax | 1,137 |
| 23 | Lightning V3.1 Pro TTS (Jun 2026) | Smallest.ai | 1,134 |
| 24 | Realtime TTS 1.5 Mini | Inworld | 1,132 |
| 25 | Azure HD 2.5 | Microsoft | 1,131 |
| 26 | Speech 2.6 Turbo | MiniMax | 1,128 |
| 27 | Fish Audio S2 Pro | Fish Audio | 1,127 |
| 28 | Async Pro v1.0 | async | 1,127 |
| 29 | Simba 3.0 | SpeechifyAI | 1,125 |
| 30 | SpaceXAI TTS | SpaceXAI | 1,125 |
| 31 | Speech-02-HD | MiniMax | 1,113 |
| 32 | TTS-1 HD | OpenAI | 1,110 |
| 33 | Step Audio EditX (Mar 2026) | StepFun | 1,104 |
| 34 | Multilingual v2 | ElevenLabs | 1,103 |
| 35 | Turbo v2.5 | ElevenLabs | 1,103 |
| 36 | TTS-1 | OpenAI | 1,097 |
| 37 | Chatterbox HD | Resemble AI | 1,093 |
| 38 | Gradium TTS | Gradium | 1,090 |
| 39 | Gemini 2.5 Flash Lite TTS | 1,088 | |
| 40 | Speech-02-Turbo | MiniMax | 1,086 |
| 41 | Studio | 1,084 | |
| 42 | Flash v2.5 | ElevenLabs | 1,083 |
| 43 | Voxtral TTS | Mistral | 1,083 |
| 44 | Speech 2.6 | Hithink | 1,081 |
| 45 | OpenAudio S1 | Fish Audio | 1,080 |
| 46 | Journey | 1,072 | |
| 47 | Sonic 3 | Cartesia | 1,071 |
| 48 | Bland Speech v3 | Bland AI | 1,071 |
| 49 | GPT-Realtime-2 | OpenAI | 1,071 |
| 50 | T2A-01-HD | MiniMax | 1,069 |
what elo means
An Elo rating from head-to-head votes: two outputs are shown side by side and people pick the better one. Higher is better, and a gap of about 100 points means the higher model wins roughly two thirds of the time.