naturalness benchmark
Direct model-level leaderboard results from Artificial Analysis.
Updated Aug 18, 2026
Higher elo ranks better on this benchmark.
Rank | Model | elo | Percentile | Participants | Evidence | Evaluated |
|---|
| Rank01 | ModelSPSimba 3.2SpeechifyAI | elo1,239 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank02 | ModelAL | elo1,237 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank03 | ModelVLLuna TTSVUI Labs | elo1,219 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank04 | ModelGO | elo1,212 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank05 | ModelST | elo1,205 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank06 | ModelCA | elo1,202 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank07 | ModelIN | elo1,199 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank08 | ModelSMLightning V3.1 Pro (Jul 2026)Smallest.ai | elo1,195 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank09 | ModelIN | elo1,187 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank10 | ModelEL | elo1,177 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank11 | ModelMI | elo1,175 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank12 | ModelMAFalcon 2Murf AI | elo1,157 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank13 | ModelGRGradium TTS BetaGradium | elo1,157 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank14 | ModelMI | elo1,153 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank15 | ModelASAsync Flash v1.5async | elo1,153 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank16 | ModelST | elo1,143 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank17 | ModelFAFish Audio S2.1 ProFish Audio | elo1,142 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank18 | ModelMI | elo1,138 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank19 | ModelSMLightning V3.1 Pro TTS (Jun 2026)Smallest.ai | elo1,138 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank20 | ModelIN | elo1,134 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank21 | ModelMI | elo1,131 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank22 | ModelASAsync Pro v1.0async | elo1,128 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank23 | ModelMI | elo1,127 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank24 | ModelFAFish Audio S2 ProFish Audio | elo1,124 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank25 | ModelSPSimba 3.0SpeechifyAI | elo1,123 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank26 | ModelMI | elo1,113 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank27 | ModelXA | elo1,113 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank28 | ModelST | elo1,110 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank29 | ModelOP | elo1,107 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank30 | ModelEL | elo1,103 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank31 | ModelEL | elo1,103 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank32 | ModelGRGradium TTSGradium | elo1,096 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank33 | ModelRAChatterbox HDResemble AI | elo1,095 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank34 | ModelOP | elo1,094 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank35 | ModelGO | elo1,088 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank36 | ModelMI | elo1,085 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank37 | ModelEL | elo1,084 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank38 | ModelGO | elo1,081 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank39 | ModelMA | elo1,081 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank40 | ModelFAOpenAudio S1Fish Audio | elo1,077 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank41 | ModelHISpeech 2.6Hithink | elo1,076 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank42 | ModelCA | elo1,072 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank43 | ModelOP | elo1,070 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank44 | ModelMRMaya 2 FlashMaya Research | elo1,069 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank45 | ModelGO | elo1,068 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank46 | ModelMI | elo1,066 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank47 | ModelNV | elo1,066 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank48 | ModelAM | elo1,064 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank49 | ModelSPSIMBA 1.6SpeechifyAI | elo1,064 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank50 | ModelXI | elo1,060 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank51 | ModelKOKokoro 82M v1.0Kokoro | elo1,059 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank52 | ModelGO | elo1,057 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank53 | ModelHAOctave 2Hume AI | elo1,057 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank54 | ModelGO | elo1,056 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank55 | ModelRI | elo1,053 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank56 | ModelASAsync Flash v1.0async | elo1,051 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank57 | ModelFAOpenAudio S1 MiniFish Audio | elo1,050 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank58 | ModelMRMaya 2 GlobalMaya Research | elo1,050 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank59 | ModelMRMaya1Maya Research | elo1,047 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank60 | ModelCA | elo1,043 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank61 | ModelBAHiggs Audio V3 TTSBoson AI | elo1,041 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank62 | ModelGO | elo1,040 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank63 | ModelAM | elo1,039 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank64 | ModelSPSIMBA 1.0SpeechifyAI | elo1,037 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank65 | ModelMI | elo1,033 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank66 | ModelMI | elo1,031 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank67 | ModelMI | elo1,031 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank68 | ModelHAOctave TTSHume AI | elo1,029 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank69 | ModelSMLightning v3.1Smallest.ai | elo1,024 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank70 | ModelRAChatterboxResemble AI | elo1,021 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank71 | ModelXI | elo1,018 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank72 | ModelNV | elo1,007 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank73 | ModelRI | elo1,003 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank74 | ModelZYZonos-v0.1Zyphra | elo1,000 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank75 | ModelMAMurf Speech Gen 2Murf AI | elo983 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank76 | ModelLMLMNTLMNT | elo975 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank77 | ModelMI | elo972 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank78 | ModelMI | elo963 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank79 | ModelOPOpenVoice v2OpenVoice | elo955 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank80 | ModelNV | elo944 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank81 | ModelAL | elo939 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank82 | ModelNENeuphonic TTSNeuphonic | elo937 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank83 | ModelAL | elo924 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank84 | ModelCOXTTS v2Coqui | elo920 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank85 | ModelGO | elo911 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank86 | ModelRI | elo897 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank87 | ModelSTStyleTTS 2StyleTTS | elo893 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank88 | ModelGO | elo890 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank89 | ModelAM | elo888 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank90 | ModelGO | elo880 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank91 | ModelNONoiz TTSNoiz | elo865 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank92 | ModelMEMetaVoice v1MetaVoice | elo844 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
| Rank93 | ModelAM | elo816 Elo | PercentileN/A | Participants93 | EvidenceA | Evaluated |
The leading models and scores on this benchmark.
A closer view of the leading scores on this benchmark.
The first five results on this benchmark, with official price and output speed added where the model identity can be matched.
Ranking basisThis artificial analysis text to speech AI model leaderboard uses descending elo in the benchmark's original unit. The leaderboard ranking keeps matched price and speed data separate from benchmark evidence.
Selection summary
Simba 3.2 currently leads Artificial Analysis Text To Speech with 1,239 Elo. It is the top model on this specific benchmark, while the best LLM for the broader task should also be checked against other benchmarks, price and runtime.
Use this leaderboard with the supporting benchmark results and coverage details above. A leaderboard position summarizes the selected ranking signal; it does not replace workload-specific testing.
What Artificial Analysis Text To Speech measures and how its scores work.
Direct model-level leaderboard results from Artificial Analysis.
Scores are shown in Elo. This benchmark is verified and has an evidence level of A.
Benchmark scores retain their original unit. Overall score eligibility is shown separately.
Common questions about Artificial Analysis Text To Speech.
Simba 3.2 is currently ranked first with 1,239 Elo.
Direct model-level leaderboard results from Artificial Analysis.
Yes. Higher values rank better for this benchmark.
93 model results are currently shown.
Yes. This benchmark can contribute to the current LLMBoard capability score.