Controlled Voice Arena LeaderboardArtificial Analysis
Compare Text to Speech (TTS) models using the same 8 cloned voices (4 US, 4 UK).
Range | Model | Elo | Samples | Released | API Pricing 1 | |
|---|---|---|---|---|---|---|
| 1 | 1-2 | HeyGen Voice | 1195±1695% interval 1179 to 1211 | 1,684 | Oct 2026 | $30.0 /1M chars |
| 2 | 2 | Qwen-Audio-3.1-TTS-Plus | 1180±1495% interval 1166 to 1194 | 2,046 | Sept 2026 | $19.3 /1M chars |
| 3 | 3-4 | Eleven v4 Turbo | 1163±1495% interval 1149 to 1177 | 2,326 | Sept 2026 | $40.0 /1M chars |
| 4 | 3-4 | Eleven v4 | 1158±1495% interval 1144 to 1172 | 2,287 | Sept 2026 | $80.0 /1M chars |
| 5 | 5-6 | Sonic 3.6 | 1143±1495% interval 1129 to 1157 | 2,900 | Aug 2026 | $49.0 /1M chars |
| 6 | 5-6 | Realtime TTS-2 | 1140±1495% interval 1126 to 1154 | 2,870 | Aug 2026 | $20.8 /1M chars |
| 7 | 7-8 | Qwen-Audio-3.0-TTS-Plus | 1125±1495% interval 1111 to 1139 | 2,615 | Sept 2026 | $19.3 /1M chars |
| 8 | 7-8 | Sonic 3.5 | 1116±1295% interval 1104 to 1128 | 4,886 | May 2026 | $49.0 /1M chars |
| 9 | 9-10 | Simba 3.2 | 1088±1495% interval 1074 to 1102 | 2,553 | Jul 2026 | $6.6 /1M chars |
| 10 | 10-12 | StepAudio 2.5 TTS | 1074±1395% interval 1061 to 1087 | 3,462 | Apr 2026 | $85.0 /1M chars |
| 11 | 10-12 | Eleven v3 | 1072±1295% interval 1060 to 1084 | 4,086 | Feb 2026 | $100.0 /1M chars |
| 12 | 10-12 | Realtime TTS-2 Flash | 1072±1395% interval 1059 to 1085 | 2,902 | Aug 2026 | $10.4 /1M chars |
| 13 | 12-18 | Gemini 3.8 Flash-Lite TTS | 1058±1495% interval 1044 to 1072 | 2,808 | Sept 2026 | $11.0 /1M chars |
| 14 | 11-18 | 1058±1595% interval 1043 to 1073 | 1,804 | Oct 2026 | $15.0 /1M chars | |
| 15 | 13-18 | v3 Conversational | 1057±1495% interval 1043 to 1071 | 2,487 | Aug 2026 | $50.0 /1M chars |
| 16 | 13-18 | SpaceXAI TTS | 1055±1395% interval 1042 to 1068 | 3,116 | Mar 2026 | $15.0 /1M chars |
| 17 | 13-18 | Gradium TTS (Sep 2026) | 1054±1495% interval 1040 to 1068 | 2,014 | Sept 2026 | $47.8 /1M chars |
| 18 | 13-18 | Gemini 3.8 Flash TTS | 1048±1495% interval 1034 to 1062 | 2,736 | Sept 2026 | $16.5 /1M chars |
| 19 | 19-21 | Simba 3.0 | 1032±1395% interval 1019 to 1045 | 2,958 | Feb 2026 | $6.6 /1M chars |
| 20 | 19-24 | Speech 2.8 HD | 1019±1395% interval 1006 to 1032 | 3,536 | Feb 2026 | $100.0 /1M chars |
| 21 | 19-24 | Gradium TTS (Aug 2026) | 1019±1495% interval 1005 to 1033 | 2,397 | Aug 2026 | $47.2 /1M chars |
| 22 | 20-24 | Speech 2.8 Turbo | 1018±1395% interval 1005 to 1031 | 3,751 | Feb 2026 | $60.0 /1M chars |
| 23 | 20-24 | 1017±1495% interval 1003 to 1031 | 2,401 | Aug 2026 | $34.0 /1M chars | |
| 24 | 20-27 | Fish Audio S2.1 Pro (Oct 2026) | 1015±1595% interval 1000 to 1030 | 1,751 | Oct 2026 | $15.0 /1M chars |
| 25 | 24-29 | 1002±1495% interval 988 to 1016 | 2,758 | Jun 2025 | $15.0 /1M chars | |
| 26 | 24-29 | 1002±1495% interval 988 to 1016 | 2,772 | Mar 2026 | $16.0 /1M chars | |
| 27 | 25-29 | 1001±1395% interval 988 to 1014 | 4,216 | Mar 2026 | $15.0 /1M chars | |
| 28 | 28 | OpenAudio S1 | 1000AnchorFixed at 1000 as the reference point for this board's Elo scale. | 5,860 | Jun 2025 | $15.0 /1M chars |
| 29 | 25-31 | Multilingual v2 | 997±1395% interval 984 to 1010 | 3,583 | Aug 2023 | $100.0 /1M chars |
| 30 | 29-31 | Fish Audio S1 | 987±1395% interval 974 to 1000 | 3,567 | Oct 2025 | $15.0 /1M chars |
| 31 | 29-31 | Flash v2.5 | 986±1395% interval 973 to 999 | 3,530 | Dec 2024 | $50.0 /1M chars |
| 32 | 31-34 | Turbo v2.5 | 973±1495% interval 959 to 987 | 2,764 | Jul 2024 | $50.0 /1M chars |
| 33 | 32-34 | Kugel 3 | 962±1395% interval 949 to 975 | 4,035 | Jun 2026 | N/A |
| 34 | 32-35 | 960±1495% interval 946 to 974 | 2,434 | Jun 2026 | $15.0 /1M chars | |
| 35 | 34-37 | Sonic 3 | 947±1495% interval 933 to 961 | 2,763 | Oct 2025 | $49.0 /1M chars |
| 36 | 35-38 | Soniox TTS Real-Time v2 | 944±1595% interval 929 to 959 | 2,244 | Aug 2026 | $14.2 /1M chars |
| 37 | 35-39 | 939±1795% interval 922 to 956 | 1,060 | Jul 2026 | N/A | |
| 38 | 36-40 | 929±1595% interval 914 to 944 | 2,720 | May 2025 | $25.0 /1M chars | |
| 39 | 37-40 | Raon SpeechLM | 928±1595% interval 913 to 943 | 2,239 | Apr 2026 | N/A |
| 40 | 38-40 | Noiz TTS | 918±1595% interval 903 to 933 | 2,303 | Nov 2025 | N/A |
| 41 | 41 | Lightning v3.1 | 886±1595% interval 871 to 901 | 2,351 | Mar 2026 | $25.0 /1M chars |
| 42 | 42 | 830±1795% interval 813 to 847 | 2,500 | Nov 2023 | $40.4 /1M chars | |
| 43 | 43 | 811±1895% interval 793 to 829 | 2,190 | Apr 2024 | $8.3 /1M chars |
1 API Pricing reflects the cost to generate 1M characters on the model creator's API with the model's default settings
2 Open Weights includes models that can be self-hosted using downloadable model weights or a provider-supplied inference runtime. Open Weights Commercial includes models whose published license permits commercial use, including use subject to license conditions (e.g. an NVIDIA AI Enterprise subscription), and excludes models that require a separately negotiated commercial agreement.