Tabla de clasificación de la Arena de voz del proveedorArtificial Analysis
Compara modelos de texto a voz (TTS) usando las voces nativas de cada proveedor.
Range | Creator | Model | Elo | 95% CI | Samples | Voces de la Arena | Lanzamiento | Precio de API 1 | |
|---|---|---|---|---|---|---|---|---|---|
| 1 | 1-2 |  Cartesia | Sonic 3.6 | 1273 | -17/17 | 1,757 | Aug 2026 | 49,0 US$ /1 M de caracteres | |
| 2 | 1-3 |  Alibaba | Qwen-Audio-3.0-TTS-Plus | 1259 | -17/17 | 1,447 | Sept 2026 | 27,6 US$ /1 M de caracteres | |
| 3 | 2-5 |  Inworld | Realtime TTS-2 | 1245 | -18/18 | 1,239 | Aug 2026 | 20,8 US$ /1 M de caracteres | |
| 4 | 3-5 |  SpeechifyAI | Simba 3.2 | 1237 | -14/14 | 2,391 | Jul 2026 | 6,6 US$ /1 M de caracteres | |
| 5 | 4-5 |  VUI Labs | Luna TTS | 1230 | -14/14 | 2,511 | Jun 2026 | 15,0 US$ /1 M de caracteres | |
| 6 | 6-10 | Inworld | Realtime TTS-2 Flash | 1210 | -15/15 | 1,691 | Aug 2026 | 10,4 US$ /1 M de caracteres | |
| 7 | 6-10 |  BreezeBlue | Breeze TTS 2  Pesos abiertos | 1204 | -16/16 | 1,390 | Aug 2026 | 34,0 US$ /1 M de caracteres | |
| 8 | 6-10 |  Google | Gemini 3.1 Flash TTS | 1199 | -12/12 | 3,430 | Apr 2026 | 18,3 US$ /1 M de caracteres | |
| 9 | 6-11 |  StepFun | StepAudio 2.5 TTS (Aug 2026) | 1199 | -17/17 | 1,279 | Aug 2026 | 85,0 US$ /1 M de caracteres | |
| 10 | 6-11 |  ElevenLabs | v3 Conversational | 1196 | -15/15 | 1,844 | Aug 2026 | 50,0 US$ /1 M de caracteres | |
| 11 | 10-12 | Cartesia | Sonic 3.5 | 1185 | -12/12 | 2,994 | May 2026 | 49,0 US$ /1 M de caracteres | |
| 12 | 11-15 |  Smallest.ai | Lightning V3.1 Pro (Jul 2026) | 1175 | -14/14 | 1,912 | Jul 2026 | 19,5 US$ /1 M de caracteres | |
| 13 | 12-15 |  Soniox | Soniox TTS Real-Time v2 | 1173 | -12/12 | 4,615 | Aug 2026 | 14,2 US$ /1 M de caracteres | |
| 14 | 12-16 |  MiniMax | Speech 2.8 HD | 1168 | -11/11 | 4,414 | Feb 2026 | 100,0 US$ /1 M de caracteres | |
| 15 | 12-16 | ElevenLabs | Eleven v3 | 1167 | -11/11 | 4,345 | Feb 2026 | 100,0 US$ /1 M de caracteres | |
| 16 | 14-18 |  Murf AI | Falcon 2 | 1157 | -15/15 | 1,740 | Jul 2026 | 10,0 US$ /1 M de caracteres | |
| 17 | 16-19 | MiniMax | Speech 2.8 Turbo | 1149 | -11/11 | 4,318 | Feb 2026 | 60,0 US$ /1 M de caracteres | |
| 18 | 16-19 |  Gradium | Gradium TTS (Aug 2026) | 1148 | -14/14 | 1,661 | Aug 2026 | 47,2 US$ /1 M de caracteres | |
| 19 | 17-24 |  Fish Audio | Fish Audio S2.1 Pro | 1139 | -13/13 | 2,162 | Jun 2026 | 15,0 US$ /1 M de caracteres | |
| 20 | 19-26 | StepFun | Step TTS 2 (Mar 2026) | 1131 | -13/13 | 2,213 | Feb 2026 | 40,0 US$ /1 M de caracteres | |
| 21 | 19-26 |  Microsoft | Azure HD 2.5 | 1128 | -12/12 | 2,405 | Mar 2026 | 22,0 US$ /1 M de caracteres | |
| 22 | 19-28 |  SpaceXAI | SpaceXAI TTS | 1128 | -15/15 | 1,210 | Mar 2026 | 15,0 US$ /1 M de caracteres | |
| 23 | 19-26 |  async | Async Flash v1.5 | 1128 | -13/13 | 2,098 | May 2026 | 10,1 US$ /1 M de caracteres | |
| 24 | 19-26 | MiniMax | Speech 2.6 HD | 1127 | -12/12 | 3,833 | Oct 2025 | 100,0 US$ /1 M de caracteres | |
| 25 | 20-29 | Fish Audio | Fish Audio S2 Pro Pesos abiertos | 1120 | -13/13 | 2,213 | Mar 2026 | 15,0 US$ /1 M de caracteres | |
| 26 | 21-29 | SpeechifyAI | Simba 3.0 | 1118 | -12/12 | 2,480 | Feb 2026 | 6,6 US$ /1 M de caracteres | |
| 27 | 25-29 | MiniMax | Speech 2.6 Turbo | 1114 | -11/11 | 4,177 | Oct 2025 | 60,0 US$ /1 M de caracteres | |
| 28 | 25-29 | async | Async Pro v1.0 | 1113 | -13/13 | 2,009 | May 2026 | 20,2 US$ /1 M de caracteres | |
| 29 | 25-30 | MiniMax | Speech-02-HD | 1111 | -12/12 | 4,246 | Mar 2025 | 100,0 US$ /1 M de caracteres | |
| 30 | 29-35 |  OpenAI | TTS-1 HD | 1099 | -12/12 | 3,473 | Nov 2023 | 30,0 US$ /1 M de caracteres | |
| 31 | 30-37 |  Resemble AI | Chatterbox HD | 1096 | -13/13 | 2,305 | — | May 2025 | 40,0 US$ /1 M de caracteres |
| 32 | 30-37 |  Hithink | Speech 2.6 | 1096 | -13/13 | 1,952 | Jul 2026 | N/D | |
| 33 | 30-37 | ElevenLabs | Turbo v2.5 | 1095 | -11/11 | 6,753 | Jul 2024 | 50,0 US$ /1 M de caracteres | |
| 34 | 30-37 | StepFun | Step Audio EditX (Mar 2026) Pesos abiertos | 1094 | -13/13 | 1,927 | Feb 2026 | N/D | |
| 35 | 30-37 | ElevenLabs | Multilingual v2 | 1092 | -10/10 | 7,199 | Aug 2023 | 100,0 US$ /1 M de caracteres | |
| 36 | 31-41 | OpenAI | TTS-1 | 1086 | -11/11 | 6,033 | Nov 2023 | 15,0 US$ /1 M de caracteres | |
| 37 | 31-43 | MiniMax | Speech-02-Turbo | 1085 | -12/12 | 3,706 | Mar 2025 | 60,0 US$ /1 M de caracteres | |
| 38 | 35-44 | Google | Gemini 2.5 Flash Lite TTS | 1081 | -12/12 | 3,064 | Dec 2025 | 9,2 US$ /1 M de caracteres | |
| 39 | 36-45 | Fish Audio | OpenAudio S1 | 1077 | -11/11 | 4,413 | Jun 2025 | 15,0 US$ /1 M de caracteres | |
| 40 | 36-45 |  Mistral | Voxtral TTS Pesos abiertos | 1076 | -13/13 | 2,036 | Mar 2026 | 16,0 US$ /1 M de caracteres | |
| 41 | 36-45 | Google | Studio | 1075 | -12/12 | 4,364 | Feb 2023 | 160,0 US$ /1 M de caracteres | |
| 42 | 37-45 | ElevenLabs | Flash v2.5 | 1074 | -11/11 | 5,605 | Dec 2024 | 50,0 US$ /1 M de caracteres | |
| 43 | 36-49 | Google | Journey | 1073 | -13/13 | 2,175 | Dec 2023 | 160,0 US$ /1 M de caracteres | |
| 44 | 37-49 | OpenAI | GPT-Realtime-2 | 1072 | -13/13 | 1,976 | May 2026 | 191,6 US$ /1 M de caracteres | |
| 45 | 39-51 |  Maya Research | Maya 2 Flash | 1066 | -14/14 | 1,819 | Jul 2026 | N/D | |
| 46 | 41-53 |  NVIDIA | Magpie-Multilingual 357M (Feb 2026) Pesos abiertos | 1063 | -13/13 | 1,971 | Feb 2026 | N/D | |
| 47 | 44-53 |  Kokoro | Kokoro 82M v1.0 Pesos abiertos | 1061 | -11/11 | 5,232 | Jan 2025 | 0,7 US$ /1 M de caracteres | |
| 48 | 45-53 | MiniMax | T2A-01-HD | 1061 | -11/11 | 4,878 | Jan 2025 | 50,0 US$ /1 M de caracteres | |
| 49 | 44-56 |  Amazon | Polly Generative | 1061 | -12/12 | 3,654 | May 2024 | 30,0 US$ /1 M de caracteres | |
| 50 | 45-57 | Cartesia | Sonic 3 | 1056 | -12/12 | 3,413 | Oct 2025 | 49,0 US$ /1 M de caracteres | |
| 51 | 45-57 |  Rime | Coda | 1054 | -13/13 | 2,188 | May 2026 | 50,0 US$ /1 M de caracteres | |
| 52 | 46-59 | Google | Chirp 3: HD | 1052 | -12/12 | 3,271 | Mar 2025 | 30,0 US$ /1 M de caracteres | |
| 53 | 46-59 | Google | Gemini 2.5 Flash TTS (Dec 2025) | 1051 | -12/12 | 2,804 | Dec 2025 | 9,2 US$ /1 M de caracteres | |
| 54 | 47-59 | async | Async Flash v1.0 | 1049 | -12/12 | 3,951 | Jul 2025 | 10,1 US$ /1 M de caracteres | |
| 55 | 48-59 |  Hume AI | Octave 2 | 1049 | -12/12 | 3,020 | Oct 2025 | 87,5 US$ /1 M de caracteres | |
| 56 | 47-59 |  Xiaomi | MiMo-V2.5-TTS | 1049 | -13/13 | 1,865 | - | N/D | |
| 57 | 50-61 | Maya Research | Maya 2 Global | 1047 | -14/14 | 1,677 | Jul 2026 | N/D | |
| 58 | 50-69 | Fish Audio | OpenAudio S1 Mini Pesos abiertos | 1041 | -20/20 | 1,681 | Jun 2025 | 15,0 US$ /1 M de caracteres | |
| 59 | 52-64 | Maya Research | Maya1 Pesos abiertos | 1041 | -12/12 | 3,205 | Jun 2025 | N/D | |
| 60 | 58-69 |  Bland AI | Bland Speech v3 | 1033 | -13/13 | 2,242 | Aug 2026 | 40,0 US$ /1 M de caracteres | |
| 61 | 58-69 |  Boson AI | Higgs Audio V3 TTS Pesos abiertos | 1033 | -13/13 | 1,970 | Jun 2026 | 15,0 US$ /1 M de caracteres | |
| 62 | 58-70 | Amazon | Polly Long-Form | 1032 | -13/13 | 1,889 | Nov 2023 | 100,0 US$ /1 M de caracteres | |
| 63 | 54-70 | Microsoft | Azure Neural | 1031 | -19/19 | 1,158 | Sept 2018 | 15,0 US$ /1 M de caracteres | |
| 64 | 58-70 | Microsoft | MAI-Voice-1 | 1030 | -13/13 | 2,134 | Apr 2026 | N/D | |
| 65 | 60-70 | Hume AI | Octave TTS | 1027 | -12/12 | 4,617 | Feb 2025 | 87,5 US$ /1 M de caracteres | |
| 66 | 60-70 | MiniMax | T2A-01-Turbo | 1027 | -11/11 | 4,905 | Jan 2025 | 30,0 US$ /1 M de caracteres | |
| 67 | 60-70 | Google | Gemini 2.5 Pro (Dec 2025) | 1027 | -13/13 | 2,403 | Dec 2025 | 18,3 US$ /1 M de caracteres | |
| 68 | 63-70 | Resemble AI | Chatterbox  Pesos abiertos | 1021 | -11/11 | 4,600 | May 2025 | 25,0 US$ /1 M de caracteres | |
| 69 | 60-70 | Xiaomi | MiMo-V2-TTS | 1021 | -14/14 | 1,716 | Mar 2026 | N/D | |
| 70 | 62-70 | Smallest.ai | Lightning v3.1 | 1020 | -13/13 | 2,089 | Mar 2026 | 25,0 US$ /1 M de caracteres | |
| 71 | 71-73 |  Krafton | Raon SpeechLM | 1002 | -14/14 | 2,028 | Apr 2026 | N/D | |
| 72 | 71-73 | Rime | Arcana v3 | 1000 | -13/13 | 2,179 | Feb 2026 | 50,0 US$ /1 M de caracteres | |
| 73 | 73 |  Zyphra | Zonos-v0.1 Pesos abiertos | 1000 | 0/0 | 4,802 | Feb 2025 | 20,0 US$ /1 M de caracteres | |
| 74 | 74 | Murf AI | Murf Speech Gen 2 | 983 | -12/12 | 4,969 | Jul 2024 | 100,0 US$ /1 M de caracteres | |
| 75 | 75-77 | Microsoft | VibeVoice 1.5B Pesos abiertos | 951 | -14/14 | 2,095 | Aug 2025 | N/D | |
| 76 | 75-78 |  OpenVoice | OpenVoice v2 Pesos abiertos | 946 | -13/13 | 2,968 | Apr 2024 | 8,3 US$ /1 M de caracteres | |
| 77 | 75-79 | Alibaba | Qwen3 TTS Flash | 941 | -14/14 | 2,117 | Sept 2025 | 10,0 US$ /1 M de caracteres | |
| 78 | 76-80 |  Neuphonic | Neuphonic TTS | 936 | -14/14 | 2,380 | Oct 2024 | 20,8 US$ /1 M de caracteres | |
| 79 | 77-80 |  KugelAudio | Kugel 3 | 931 | -13/13 | 4,132 | Jun 2026 | N/D | |
| 80 | 78-82 | Alibaba | Qwen3 TTS | 926 | -13/13 | 2,861 | Jan 2026 | 20,0 US$ /1 M de caracteres | |
| 81 | 80-83 | Google | WaveNet | 915 | -12/12 | 6,225 | Sept 2016 | 4,0 US$ /1 M de caracteres | |
| 82 | 80-83 |  Coqui | XTTS v2 Pesos abiertos | 914 | -15/15 | 2,523 | — | Nov 2023 | 40,4 US$ /1 M de caracteres |
| 83 | 81-85 | Rime | Mist V2 | 906 | -15/15 | 2,825 | Mar 2025 | 31,1 US$ /1 M de caracteres | |
| 84 | 84-86 | Google | Neural2 | 894 | -12/12 | 6,455 | Jun 2022 | 16,0 US$ /1 M de caracteres | |
| 85 | 83-88 |  StyleTTS | StyleTTS 2 Pesos abiertos | 891 | -15/15 | 2,437 | — | Jun 2023 | 2,8 US$ /1 M de caracteres |
| 86 | 84-88 | Amazon | Polly Neural | 890 | -14/14 | 2,735 | Jul 2019 | 16,0 US$ /1 M de caracteres | |
| 87 | 85-88 | Google | Standard | 881 | -12/12 | 6,491 | Aug 2018 | 4,0 US$ /1 M de caracteres | |
| 88 | 84-88 |  Noiz | Noiz TTS | 881 | -15/15 | 1,958 | Nov 2025 | N/D | |
| 89 | 89-90 |  MetaVoice | MetaVoice v1 Pesos abiertos | 845 | -18/18 | 1,349 | — | Feb 2024 | N/D |
| 90 | 90 | Amazon | Polly Standard | 828 | -16/16 | 3,286 | Nov 2016 | 4,0 US$ /1 M de caracteres |
1 El precio de API refleja el costo de generar 1M de caracteres en la API del creador del modelo con la configuración predeterminada del modelo
2 Pesos abiertos incluye modelos que pueden autoalojarse mediante pesos descargables o un entorno de inferencia proporcionado por el proveedor. Pesos abiertos comerciales incluye modelos cuya licencia publicada permite el uso comercial, incluso sujeto a condiciones de la licencia (p. ej., una suscripción a NVIDIA AI Enterprise), y excluye los modelos que requieren un acuerdo comercial negociado por separado.
Preguntas frecuentes
Los mejores modelos de texto a voz por puntaje Elo son: 1. Sonic 3.6 (Elo 1273), 2. Qwen-Audio-3.0-TTS-Plus (Elo 1259), 3. Realtime TTS-2 (Elo 1245), 4. Simba 3.2 (Elo 1237) y 5. Luna TTS (Elo 1230). Los rankings se basan en votos a ciegas de usuarios en la Arena de voz.
Kokoro 82M v1.0 es el más asequible a $0.65 por 1M de caracteres, con un puntaje Elo de 1061. Otras opciones asequibles incluyen StyleTTS 2 a $2.82 por 1M de caracteres.
Los 5 mejores modelos de texto a voz de pesos abiertos son: 1. Breeze TTS 2 (Elo 1204), 2. Fish Audio S2 Pro (Elo 1120), 3. Step Audio EditX (Mar 2026) (Elo 1094), 4. Voxtral TTS (Elo 1076) y 5. Magpie-Multilingual 357M (Feb 2026) (Elo 1063).