Ranking da Arena de voz do provedorArtificial Analysis
Compare modelos de texto para fala (TTS) usando as vozes nativas de cada provedor.
Faixa | Modelo | Elo | Amostras | Vozes da Arena | Lançamento | Preço da API 1 | |
|---|---|---|---|---|---|---|---|
| 1 | 1-2 |  Eleven v4 Turbo | 1327±18Intervalo de 95%: 1309 a 1345 | 1,768 | Sept 2026 | US$ 40,0 /1 mi de caracteres | |
| 2 | 1-2 | Eleven v4 | 1323±17Intervalo de 95%: 1306 a 1340 | 2,057 | Sept 2026 | US$ 80,0 /1 mi de caracteres | |
| 3 | 3-4 |  Qwen-Audio-3.1-TTS-Plus | 1298±17Intervalo de 95%: 1281 a 1315 | 1,651 | Sept 2026 | US$ 19,3 /1 mi de caracteres | |
| 4 | 3-5 |  Sonic 3.6 | 1284±15Intervalo de 95%: 1269 a 1299 | 2,435 | Aug 2026 | US$ 49,0 /1 mi de caracteres | |
| 5 | 4-6 |  Gemini 3.8 Flash TTS | 1272±15Intervalo de 95%: 1257 a 1287 | 2,677 | Sept 2026 | US$ 16,5 /1 mi de caracteres | |
| 6 | 5-7 | Qwen-Audio-3.0-TTS-Plus | 1264±15Intervalo de 95%: 1249 a 1279 | 2,123 | Sept 2026 | US$ 19,3 /1 mi de caracteres | |
| 7 | 5-9 |  Realtime TTS-2 | 1257±16Intervalo de 95%: 1241 a 1273 | 1,884 | Aug 2026 | US$ 20,8 /1 mi de caracteres | |
| 8 | 7-9 | Gemini 3.8 Flash-Lite TTS | 1244±14Intervalo de 95%: 1230 a 1258 | 2,639 | Sept 2026 | US$ 11,0 /1 mi de caracteres | |
| 9 | 8-9 |  Simba 3.2 | 1243±13Intervalo de 95%: 1230 a 1256 | 3,017 | Jul 2026 | US$ 6,6 /1 mi de caracteres | |
| 10 | 10-15 |  Breeze TTS 2  | 1222±15Intervalo de 95%: 1207 a 1237 | 1,940 | Aug 2026 | US$ 34,0 /1 mi de caracteres | |
| 11 | 10-15 |  Luna TTS | 1218±12Intervalo de 95%: 1206 a 1230 | 3,185 | Jun 2026 | US$ 15,0 /1 mi de caracteres | |
| 12 | 10-15 | Realtime TTS-2 Flash | 1217±14Intervalo de 95%: 1203 a 1231 | 2,301 | Aug 2026 | US$ 10,4 /1 mi de caracteres | |
| 13 | 10-15 | Gemini 3.1 Flash TTS | 1212±11Intervalo de 95%: 1201 a 1223 | 3,971 | Apr 2026 | US$ 18,3 /1 mi de caracteres | |
| 14 | 10-15 |  StepAudio 2.5 TTS (Aug 2026) | 1209±15Intervalo de 95%: 1194 a 1224 | 1,843 | Aug 2026 | US$ 85,0 /1 mi de caracteres | |
| 15 | 11-15 | v3 Conversational | 1207±14Intervalo de 95%: 1193 a 1221 | 2,401 | Aug 2026 | US$ 50,0 /1 mi de caracteres | |
| 16 | 16-17 | Sonic 3.5 | 1191±12Intervalo de 95%: 1179 a 1203 | 3,464 | May 2026 | US$ 49,0 /1 mi de caracteres | |
| 17 | 17-20 |  Soniox TTS Real-Time v2 | 1179±11Intervalo de 95%: 1168 a 1190 | 5,066 | Aug 2026 | US$ 14,2 /1 mi de caracteres | |
| 18 | 17-21 | Eleven v3 | 1176±11Intervalo de 95%: 1165 a 1187 | 4,766 | Feb 2026 | US$ 100,0 /1 mi de caracteres | |
| 19 | 17-21 |  Speech 2.8 HD | 1173±11Intervalo de 95%: 1162 a 1184 | 4,848 | Feb 2026 | US$ 100,0 /1 mi de caracteres | |
| 20 | 17-21 |  Lightning V3.1 Pro (Jul 2026) | 1172±13Intervalo de 95%: 1159 a 1185 | 2,405 | Jul 2026 | US$ 19,5 /1 mi de caracteres | |
| 21 | 17-21 |  Gradium TTS (Sep 2026) | 1168±14Intervalo de 95%: 1154 a 1182 | 1,814 | Sept 2026 | US$ 47,8 /1 mi de caracteres | |
| 22 | 22-25 | Speech 2.8 Turbo | 1153±11Intervalo de 95%: 1142 a 1164 | 4,667 | Feb 2026 | US$ 60,0 /1 mi de caracteres | |
| 23 | 22-26 | Gradium TTS (Aug 2026) | 1152±13Intervalo de 95%: 1139 a 1165 | 2,070 | Aug 2026 | US$ 47,2 /1 mi de caracteres | |
| 24 | 22-26 |  Falcon 2 | 1151±14Intervalo de 95%: 1137 a 1165 | 2,123 | Jul 2026 | US$ 10,0 /1 mi de caracteres | |
| 25 | 22-28 |  SpaceXAI TTS | 1143±14Intervalo de 95%: 1129 a 1157 | 1,536 | Mar 2026 | US$ 15,0 /1 mi de caracteres | |
| 26 | 22-28 |  Fish Audio S2.1 Pro | 1141±12Intervalo de 95%: 1129 a 1153 | 2,518 | Jun 2026 | US$ 15,0 /1 mi de caracteres | |
| 27 | 25-30 | Step TTS 2 (Mar 2026) | 1137±12Intervalo de 95%: 1125 a 1149 | 2,512 | Feb 2026 | US$ 40,0 /1 mi de caracteres | |
| 28 | 25-30 |  Async Flash v1.5 | 1136±12Intervalo de 95%: 1124 a 1148 | 2,385 | May 2026 | US$ 10,1 /1 mi de caracteres | |
| 29 | 27-33 |  Azure HD 2.5 | 1129±12Intervalo de 95%: 1117 a 1141 | 2,680 | Mar 2026 | US$ 22,0 /1 mi de caracteres | |
| 30 | 27-33 | Speech 2.6 HD | 1128±11Intervalo de 95%: 1117 a 1139 | 4,039 | Oct 2025 | US$ 100,0 /1 mi de caracteres | |
| 31 | 29-35 | Speech 2.6 Turbo | 1123±11Intervalo de 95%: 1112 a 1134 | 4,355 | Oct 2025 | US$ 60,0 /1 mi de caracteres | |
| 32 | 29-35 | Simba 3.0 | 1123±12Intervalo de 95%: 1111 a 1135 | 2,743 | Feb 2026 | US$ 6,6 /1 mi de caracteres | |
| 33 | 29-35 | Async Pro v1.0 | 1122±12Intervalo de 95%: 1110 a 1134 | 2,279 | May 2026 | US$ 20,2 /1 mi de caracteres | |
| 34 | 30-36 | Fish Audio S2 Pro | 1116±12Intervalo de 95%: 1104 a 1128 | 2,483 | Mar 2026 | US$ 15,0 /1 mi de caracteres | |
| 35 | 31-37 | Speech-02-HD | 1113±11Intervalo de 95%: 1102 a 1124 | 4,442 | Mar 2025 | US$ 100,0 /1 mi de caracteres | |
| 36 | 34-41 |  TTS-1 HD | 1105±11Intervalo de 95%: 1094 a 1116 | 3,613 | Nov 2023 | US$ 30,0 /1 mi de caracteres | |
| 37 | 35-41 |  Chatterbox HD | 1103±12Intervalo de 95%: 1091 a 1115 | 2,471 | — | May 2025 | US$ 40,0 /1 mi de caracteres |
| 38 | 35-43 |  Speech 2.6 | 1100±13Intervalo de 95%: 1087 a 1113 | 2,187 | Jul 2026 | N/D | |
| 39 | 36-43 | Turbo v2.5 | 1099±10Intervalo de 95%: 1089 a 1109 | 6,876 | Jul 2024 | US$ 50,0 /1 mi de caracteres | |
| 40 | 36-45 | Step Audio EditX (Mar 2026) | 1096±13Intervalo de 95%: 1083 a 1109 | 2,130 | Feb 2026 | N/D | |
| 41 | 37-44 | Multilingual v2 | 1095±10Intervalo de 95%: 1085 a 1105 | 7,317 | Aug 2023 | US$ 100,0 /1 mi de caracteres | |
| 42 | 38-47 | Speech-02-Turbo | 1091±12Intervalo de 95%: 1079 a 1103 | 3,819 | Mar 2025 | US$ 60,0 /1 mi de caracteres | |
| 43 | 38-47 | TTS-1 | 1089±11Intervalo de 95%: 1078 a 1100 | 6,149 | Nov 2023 | US$ 15,0 /1 mi de caracteres | |
| 44 | 40-49 |  Voxtral TTS | 1086±13Intervalo de 95%: 1073 a 1099 | 2,214 | Mar 2026 | US$ 16,0 /1 mi de caracteres | |
| 45 | 40-50 | Gemini 2.5 Flash Lite TTS | 1084±12Intervalo de 95%: 1072 a 1096 | 3,163 | Dec 2025 | US$ 9,2 /1 mi de caracteres | |
| 46 | 42-50 | OpenAudio S1 | 1082±11Intervalo de 95%: 1071 a 1093 | 4,505 | Jun 2025 | US$ 15,0 /1 mi de caracteres | |
| 47 | 42-50 | Studio | 1081±11Intervalo de 95%: 1070 a 1092 | 4,456 | Feb 2023 | US$ 160,0 /1 mi de caracteres | |
| 48 | 42-53 | GPT-Realtime-2 | 1078±13Intervalo de 95%: 1065 a 1091 | 2,112 | May 2026 | US$ 191,6 /1 mi de caracteres | |
| 49 | 44-50 | Flash v2.5 | 1077±11Intervalo de 95%: 1066 a 1088 | 5,700 | Dec 2024 | US$ 50,0 /1 mi de caracteres | |
| 50 | 45-55 | Journey | 1073±13Intervalo de 95%: 1060 a 1086 | 2,236 | Dec 2023 | US$ 160,0 /1 mi de caracteres | |
| 51 | 48-61 |  Maya 2 Flash | 1066±13Intervalo de 95%: 1053 a 1079 | 1,957 | Jul 2026 | N/D | |
| 52 | 50-59 | T2A-01-HD | 1066±11Intervalo de 95%: 1055 a 1077 | 4,953 | Jan 2025 | US$ 50,0 /1 mi de caracteres | |
| 53 | 49-61 |  Polly Generative | 1066±12Intervalo de 95%: 1054 a 1078 | 3,728 | May 2024 | US$ 30,0 /1 mi de caracteres | |
| 54 | 50-61 |  Kokoro 82M v1.0 | 1064±11Intervalo de 95%: 1053 a 1075 | 5,307 | Jan 2025 | US$ 0,7 /1 mi de caracteres | |
| 55 | 50-63 |  Magpie-Multilingual 357M (Feb 2026) | 1062±13Intervalo de 95%: 1049 a 1075 | 2,129 | Feb 2026 | N/D | |
| 56 | 51-63 | Sonic 3 | 1060±12Intervalo de 95%: 1048 a 1072 | 3,477 | Oct 2025 | US$ 49,0 /1 mi de caracteres | |
| 57 | 51-63 | Chirp 3: HD | 1059±12Intervalo de 95%: 1047 a 1071 | 3,340 | Mar 2025 | US$ 30,0 /1 mi de caracteres | |
| 58 | 51-65 | Gemini 2.5 Flash TTS (Dec 2025) | 1056±12Intervalo de 95%: 1044 a 1068 | 2,856 | Dec 2025 | US$ 9,2 /1 mi de caracteres | |
| 59 | 51-65 |  Coda | 1056±12Intervalo de 95%: 1044 a 1068 | 2,300 | May 2026 | US$ 50,0 /1 mi de caracteres | |
| 60 | 51-66 |  MiMo-V2.5-TTS | 1055±13Intervalo de 95%: 1042 a 1068 | 1,993 | - | N/D | |
| 61 | 51-66 |  Flux TTS | 1054±14Intervalo de 95%: 1040 a 1068 | 2,012 | Aug 2026 | US$ 45,0 /1 mi de caracteres | |
| 62 | 55-66 | Async Flash v1.0 | 1051±11Intervalo de 95%: 1040 a 1062 | 4,006 | Jul 2025 | US$ 10,1 /1 mi de caracteres | |
| 63 | 55-67 |  Octave 2 | 1051±12Intervalo de 95%: 1039 a 1063 | 3,089 | Oct 2025 | US$ 87,5 /1 mi de caracteres | |
| 64 | 58-70 | Maya1 | 1046±12Intervalo de 95%: 1034 a 1058 | 3,251 | Jun 2025 | N/D | |
| 65 | 54-76 | OpenAudio S1 Mini | 1044±20Intervalo de 95%: 1024 a 1064 | 1,682 | Jun 2025 | US$ 15,0 /1 mi de caracteres | |
| 66 | 58-73 | Maya 2 Global | 1043±14Intervalo de 95%: 1029 a 1057 | 1,849 | Jul 2026 | N/D | |
| 67 | 64-73 | Octave TTS | 1039±11Intervalo de 95%: 1028 a 1050 | 4,656 | Feb 2025 | US$ 87,5 /1 mi de caracteres | |
| 68 | 64-76 |  Higgs Audio V3 TTS | 1036±13Intervalo de 95%: 1023 a 1049 | 2,102 | Jun 2026 | US$ 15,0 /1 mi de caracteres | |
| 69 | 62-77 | Azure Neural | 1035±19Intervalo de 95%: 1016 a 1054 | 1,158 | Sept 2018 | US$ 15,0 /1 mi de caracteres | |
| 70 | 64-76 |  Bland Speech v3 | 1035±13Intervalo de 95%: 1022 a 1048 | 2,380 | Aug 2026 | US$ 40,0 /1 mi de caracteres | |
| 71 | 66-77 | MAI-Voice-1 | 1031±13Intervalo de 95%: 1018 a 1044 | 2,261 | Apr 2026 | N/D | |
| 72 | 66-77 | Gemini 2.5 Pro (Dec 2025) | 1030±13Intervalo de 95%: 1017 a 1043 | 2,435 | Dec 2025 | US$ 18,3 /1 mi de caracteres | |
| 73 | 67-77 | Polly Long-Form | 1030±13Intervalo de 95%: 1017 a 1043 | 1,927 | Nov 2023 | US$ 100,0 /1 mi de caracteres | |
| 74 | 68-77 | T2A-01-Turbo | 1028±11Intervalo de 95%: 1017 a 1039 | 4,944 | Jan 2025 | US$ 30,0 /1 mi de caracteres | |
| 75 | 68-77 | Chatterbox  | 1026±11Intervalo de 95%: 1015 a 1037 | 4,646 | May 2025 | US$ 25,0 /1 mi de caracteres | |
| 76 | 68-77 | Lightning v3.1 | 1024±13Intervalo de 95%: 1011 a 1037 | 2,218 | Mar 2026 | US$ 25,0 /1 mi de caracteres | |
| 77 | 69-77 | MiMo-V2-TTS | 1022±14Intervalo de 95%: 1008 a 1036 | 1,805 | Mar 2026 | N/D | |
| 78 | 78-79 |  Raon SpeechLM | 1002±13Intervalo de 95%: 989 a 1015 | 2,116 | Apr 2026 | N/D | |
| 79 | 79 |  Zonos-v0.1 | 1000ÂncoraFixado em 1000 como ponto de referência da escala Elo deste ranking. | 4,839 | Feb 2025 | US$ 20,0 /1 mi de caracteres | |
| 80 | 80 | Murf Speech Gen 2 | 985±12Intervalo de 95%: 973 a 997 | 4,988 | Jul 2024 | US$ 100,0 /1 mi de caracteres | |
| 81 | 81-82 | VibeVoice 1.5B | 957±14Intervalo de 95%: 943 a 971 | 2,123 | Aug 2025 | N/D | |
| 82 | 81-83 |  OpenVoice v2 | 955±14Intervalo de 95%: 941 a 969 | 3,010 | Apr 2024 | US$ 8,3 /1 mi de caracteres | |
| 83 | 82-86 | Qwen3 TTS Flash | 942±14Intervalo de 95%: 928 a 956 | 2,160 | Sept 2025 | US$ 11,0 /1 mi de caracteres | |
| 84 | 83-86 |  Neuphonic TTS | 940±14Intervalo de 95%: 926 a 954 | 2,413 | Oct 2024 | US$ 20,8 /1 mi de caracteres | |
| 85 | 83-86 |  Kugel 3 | 933±13Intervalo de 95%: 920 a 946 | 4,250 | Jun 2026 | N/D | |
| 86 | 83-86 | Qwen3 TTS | 932±13Intervalo de 95%: 919 a 945 | 2,979 | Jan 2026 | US$ 13,8 /1 mi de caracteres | |
| 87 | 87-89 | WaveNet | 917±12Intervalo de 95%: 905 a 929 | 6,300 | Sept 2016 | US$ 4,0 /1 mi de caracteres | |
| 88 | 87-89 |  XTTS v2 | 916±15Intervalo de 95%: 901 a 931 | 2,592 | — | Nov 2023 | US$ 40,4 /1 mi de caracteres |
| 89 | 87-92 | Mist V2 | 907±15Intervalo de 95%: 892 a 922 | 2,924 | Mar 2025 | US$ 31,1 /1 mi de caracteres | |
| 90 | 89-93 | Polly Neural | 896±14Intervalo de 95%: 882 a 910 | 2,873 | Jul 2019 | US$ 16,0 /1 mi de caracteres | |
| 91 | 90-93 | Neural2 | 895±12Intervalo de 95%: 883 a 907 | 6,578 | Jun 2022 | US$ 16,0 /1 mi de caracteres | |
| 92 | 89-93 |  StyleTTS 2 | 894±15Intervalo de 95%: 879 a 909 | 2,550 | — | Jun 2023 | US$ 2,8 /1 mi de caracteres |
| 93 | 90-94 | Standard | 887±12Intervalo de 95%: 875 a 899 | 6,637 | Aug 2018 | US$ 4,0 /1 mi de caracteres | |
| 94 | 93-94 |  Noiz TTS | 877±15Intervalo de 95%: 862 a 892 | 2,229 | Nov 2025 | N/D | |
| 95 | 95-96 |  MetaVoice v1 | 848±18Intervalo de 95%: 830 a 866 | 1,546 | — | Feb 2024 | N/D |
| 96 | 96 | Polly Standard | 831±16Intervalo de 95%: 815 a 847 | 3,489 | Nov 2016 | US$ 4,0 /1 mi de caracteres |
1 O preço da API reflete o custo de gerar 1M de caracteres na API do criador do modelo com as configurações padrão do modelo
2 Pesos abertos inclui modelos que podem ser auto-hospedados usando pesos de modelo para download ou um ambiente de inferência fornecido pelo provedor. Pesos abertos para uso comercial inclui modelos cuja licença publicada permite o uso comercial, inclusive sujeito a condições da licença (por exemplo, uma assinatura do NVIDIA AI Enterprise), e exclui modelos que exigem um acordo comercial negociado separadamente.
Perguntas frequentes
Os melhores modelos de texto para fala por classificação Elo são: 1. Eleven v4 Turbo (Elo 1327), 2. Eleven v4 (Elo 1323), 3. Qwen-Audio-3.1-TTS-Plus (Elo 1298), 4. Sonic 3.6 (Elo 1284) e 5. Gemini 3.8 Flash TTS (Elo 1272). Os rankings se baseiam em votos cegos de usuários na Speech Arena.
Kokoro 82M v1.0 é o mais acessível a $0.65 por 1M de caracteres, com uma pontuação Elo de 1064. Outras opções acessíveis incluem StyleTTS 2 a $2.82 por 1M de caracteres.
Os 5 melhores modelos de texto para fala de pesos abertos são: 1. Breeze TTS 2 (Elo 1222), 2. Fish Audio S2 Pro (Elo 1116), 3. Step Audio EditX (Mar 2026) (Elo 1096), 4. Voxtral TTS (Elo 1086) e 5. Kokoro 82M v1.0 (Elo 1064).