Benchmark Almanac

Voice Arena US English

Gemini 3.8 Flash-Lite TTS 1,081 leads since September 2026
Gemini 3.8 Flash-Lite TTS 1,087 leads since September 2026
Only in system cards

Models generate speech in US English. Listeners compare anonymous clips, and preference votes determine independent language Elo ratings.

Each language has an independent leaderboard. Topic filters stay within its vote pool.

Made by
Voice Arena
Kind
Audio understanding and voice
Level
Everyday
Scored
Ranked by votes
Answers
Open-ended
Tasks
All public
Last in a system card
September 2026
Last model added
October 2026
Scores refreshed
November 12th, 2026
25
Entries on its own leaderboard, 7 models we track
2
System cards report it, from 1 lab

Gemini 3.8 Flash-Lite TTS leads with 1,081

Best score per entry from the benchmark's own leaderboard, read on November 12th, 2026. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders.

Gemini 3.8 Flash-Lite TTS leads with 1,087, claims included

Best score per model from every source, including labs’ own numbers and claims Epoch AI collected from reports and papers. Labs pick their own settings, so a claim can beat an independent run of the same model.

25 entries. Scroll sideways for the rest. Bars start at 850.

  1. 1,081 Gemini 3.8 Flash-Lite TTS
  2. 1,070 ElevenLabs Eleven v4
  3. 1,063 Gemini 3.8 Flash TTS
  4. 1,061 Cartesia Sonic-3.6
  5. 1,056 Gemini 3.1 Flash TTS
  6. 1,053 Inworld Realtime TTS 2 Research Preview
  7. 1,050 Smallest AI Lightning 3.1 Pro (October)
  8. 1,032 SpeechifyAI Simba 3.2
  9. 1,031 Cartesia Sonic-3.5
  10. 1,024 ElevenLabs Eleven v3 Conversational
  11. 1,013 Voxtral Mini TTS 3
  12. 1,010 Gradium-TTS
  13. 998 Fish Audio Drama 3
  14. 995 Grok TTS
  15. 991 Fish Audio S2.1 Pro
  16. 991 Murf AI Falcon 2
  17. 986 ElevenLabs Eleven v3
  18. 985 Maya Research Maya-2-Global
  19. 974 Dragon HD Omni
  20. 967 Hithink Speech 2.6
  21. 959 Smallest AI Lightning 3.1 Pro
  22. 939 Fish Audio S2 Pro
  23. 928 GPT-4o Mini TTS
  24. 899 SILMA TTS v2
  25. 859 Cartesia Sonic-3

Bars start at 850.

  1. 1 Gemini 3.8 Flash-Lite TTS 1,081
  2. 2 ElevenLabs Eleven v4 1,070
  3. 3 Gemini 3.8 Flash TTS 1,063
  4. 4 Cartesia Sonic-3.6 1,061
  5. 5 Gemini 3.1 Flash TTS 1,056
  6. 6 Inworld Realtime TTS 2 Research Preview 1,053
  7. 7 Smallest AI Lightning 3.1 Pro (October) 1,050
  8. 8 SpeechifyAI Simba 3.2 1,032
  9. 9 Cartesia Sonic-3.5 1,031
  10. 10 ElevenLabs Eleven v3 Conversational 1,024
  11. 11 Voxtral Mini TTS 3 1,013
  12. 12 Gradium-TTS 1,010
  13. 13 Fish Audio Drama 3 998
  14. 14 Grok TTS 995
  15. 15 Fish Audio S2.1 Pro 991
  16. 16 Murf AI Falcon 2 991
  17. 17 ElevenLabs Eleven v3 986
  18. 18 Maya Research Maya-2-Global 985
  19. 19 Dragon HD Omni 974
  20. 20 Hithink Speech 2.6 967
  21. 21 Smallest AI Lightning 3.1 Pro 959
  22. 22 Fish Audio S2 Pro 939
  23. 23 GPT-4o Mini TTS 928
  24. 24 SILMA TTS v2 899
  25. 25 Cartesia Sonic-3 859

Source: Official Voice Arena US English leaderboard

Sources: Labs’ model cards and official Voice Arena US English leaderboard

#EntryScoreSettingDateSource
1 Gemini 3.8 Flash-Lite TTS 1,081 Sep 2026 Official leaderboard
2 ElevenLabs Eleven v4 1,070 Official leaderboard
3 Gemini 3.8 Flash TTS 1,063 Sep 2026 Official leaderboard
4 Cartesia Sonic-3.6 1,061 Official leaderboard
5 Gemini 3.1 Flash TTS 1,056 Apr 2026 Official leaderboard
6 Inworld Realtime TTS 2 Research Preview 1,053 Official leaderboard
7 Smallest AI Lightning 3.1 Pro (October) 1,050 Official leaderboard
8 SpeechifyAI Simba 3.2 1,032 Official leaderboard
9 Cartesia Sonic-3.5 1,031 Official leaderboard
10 ElevenLabs Eleven v3 Conversational 1,024 Official leaderboard
11 Voxtral Mini TTS 3 1,013 Oct 2026 Official leaderboard
12 Gradium-TTS 1,010 Official leaderboard
13 Fish Audio Drama 3 998 Official leaderboard
14 Grok TTS 995 Apr 2026 Official leaderboard
15 Fish Audio S2.1 Pro 991 Official leaderboard
16 Murf AI Falcon 2 991 Official leaderboard
17 ElevenLabs Eleven v3 986 Official leaderboard
18 Maya Research Maya-2-Global 985 Official leaderboard
19 Dragon HD Omni 974 Dec 2025 Official leaderboard
20 Hithink Speech 2.6 967 Official leaderboard
21 Smallest AI Lightning 3.1 Pro 959 Official leaderboard
22 Fish Audio S2 Pro 939 Official leaderboard
23 GPT-4o Mini TTS 928 Mar 2025 Official leaderboard
24 SILMA TTS v2 899 Official leaderboard
25 Cartesia Sonic-3 859 Official leaderboard
#ModelScoreSettingDateSource
1 Gemini 3.8 Flash-Lite TTS 1,087 Sep 2026 Lab model card
2 Gemini 3.8 Flash TTS 1,063 Sep 2026 Official leaderboard
3 Gemini 3.1 Flash TTS 1,056 Apr 2026 Official leaderboard
4 Voxtral Mini TTS 3 1,013 Oct 2026 Official leaderboard
5 Grok TTS 995 Apr 2026 Official leaderboard
6 Dragon HD Omni 974 Dec 2025 Official leaderboard
7 GPT-4o Mini TTS 928 Mar 2025 Official leaderboard

Labs report scores for 2 models in their cards. Switch the scores above to rank them together.

Quality vs Speed

Published US English latency summaries use the default 1-day measurement window. The line joins the runs that no other run beats on both measures. We redrew this chart from the numbers on the official leaderboard.

900 1,000 1,100 1,200 02004006008001,000 Median time to first audio (ms), lower is better Elo GPT-4o Mini TTS: Elo 928, median time to first audio 812 ms Dragon HD Omni: Elo 974, median time to first audio 485 ms Gemini 3.1 Flash TTS: Elo 1,056, median time to first audio 890 ms Grok TTS: Elo 995, median time to first audio 354 ms Cartesia Sonic-3.5: Elo 1,031, median time to first audio 250 ms. Not in our model list Fish Audio S2.1 Pro: Elo 991, median time to first audio 283 ms. Not in our model list Murf AI Falcon 2: Elo 991, median time to first audio 102 ms. Not in our model list ElevenLabs Eleven v3: Elo 986, median time to first audio 588 ms. Not in our model list Smallest AI Lightning 3.1 Pro: Elo 959, median time to first audio 264 ms. Not in our model list Fish Audio S2 Pro: Elo 939, median time to first audio 267 ms. Not in our model list Cartesia Sonic-3.6: Elo 1,061, median time to first audio 341 ms. Not in our model list Inworld Realtime TTS 2 Research Preview: Elo 1,053, median time to first audio 169 ms. Not in our model list SpeechifyAI Simba 3.2: Elo 1,032, median time to first audio 123 ms. Not in our model list Gradium-TTS: Elo 1,010, median time to first audio 68 ms. Not in our model list Cartesia Sonic-3.6Inworld Realtime TTS 2 Research PreviewSpeechifyAI Simba 3.2Gradium-TTS 900 1,000 1,100 1,200 02004006008001,000 Median time to first audio (ms), lower is better Elo GPT-4o Mini TTS: Elo 928, median time to first audio 812 ms Dragon HD Omni: Elo 974, median time to first audio 485 ms Gemini 3.1 Flash TTS: Elo 1,056, median time to first audio 890 ms Grok TTS: Elo 995, median time to first audio 354 ms Cartesia Sonic-3.5: Elo 1,031, median time to first audio 250 ms. Not in our model list Fish Audio S2.1 Pro: Elo 991, median time to first audio 283 ms. Not in our model list Murf AI Falcon 2: Elo 991, median time to first audio 102 ms. Not in our model list ElevenLabs Eleven v3: Elo 986, median time to first audio 588 ms. Not in our model list Smallest AI Lightning 3.1 Pro: Elo 959, median time to first audio 264 ms. Not in our model list Fish Audio S2 Pro: Elo 939, median time to first audio 267 ms. Not in our model list Cartesia Sonic-3.6: Elo 1,061, median time to first audio 341 ms. Not in our model list Inworld Realtime TTS 2 Research Preview: Elo 1,053, median time to first audio 169 ms. Not in our model list SpeechifyAI Simba 3.2: Elo 1,032, median time to first audio 123 ms. Not in our model list Gradium-TTS: Elo 1,010, median time to first audio 68 ms. Not in our model list Cartesia Sonic-3.6Inworld Realtime TTS 2 Research PreviewSpeechifyAI Simba 3.2Gradium-TTS
ModelSettingEloMedian time to first audio
Cartesia Sonic-3.6 1,061 340.5 ms
Gemini 3.1 Flash TTS 1,056 890 ms
Inworld Realtime TTS 2 Research Preview 1,053 168.5 ms
SpeechifyAI Simba 3.2 1,032 123 ms
Cartesia Sonic-3.5 1,031 250 ms
Gradium-TTS 1,010 68 ms
Grok TTS 995 354 ms
Murf AI Falcon 2 991 102 ms
Fish Audio S2.1 Pro 991 283 ms
ElevenLabs Eleven v3 986 587.5 ms
Dragon HD Omni 974 485 ms
Smallest AI Lightning 3.1 Pro 959 263.5 ms
Fish Audio S2 Pro 939 267 ms
GPT-4o Mini TTS 928 811.5 ms

4 models have held number one on the official leaderboard since 2023 4 models have held number one with claims included since 2023

Lab claims count from the day the card came out, independent runs from the model’s release.

OpenAI Microsoft Google DeepMind
Own leaderboard 4 leaders
Gemini 3.8 Flash-Lite TTS 1,081

Source: Official Voice Arena US English leaderboard

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

On the benchmark’s own leaderboard Reported by the lab in its model card
900 950 1,000 1,050 1,100 2025 2026 2027 900 950 1,000 1,050 1,100 2025 2026 2027

Sources: Official Voice Arena US English leaderboard and labs’ model cards

Labs report it only in system cards

We read the headline results of 522 launch posts from 52 labs, and none of them list this benchmark. Some system cards do, below.

2 system cards report it, from 1 lab

Besides launch posts, we read the system cards, model cards and technical reports labs publish with a release. They report far more, including safety and frontier-risk tests.

DateDocumentScoreAs the document writes itSection
Sep 23, 2026 Gemini 3.8 Flash TTS model card 1061 Voice Arena us english, Elo, Gemini speech generation model page Capabilities
Sep 23, 2026 Gemini 3.8 Flash-Lite TTS model card 1087 Voice Arena us english, Elo, Gemini speech generation model page Capabilities

2 lab claims next to the leaderboard

Each score a lab reported for its own model, beside the same model's best score on the leaderboard. Gaps mostly come from different harnesses, effort levels and task subsets. Every benchmark.

Model Lab says Board says Gap
Gemini 3.8 Flash-Lite TTS 1,087 Voice Arena us english, Elo, Gemini speech generation model page 1,081 Leaderboard +6
Gemini 3.8 Flash TTS 1,061 Voice Arena us english, Elo, Gemini speech generation model page 1,063 Leaderboard −2

More in Audio and video

All 233 in Audio and video
  • Video-MMMU In use 300 expert-level lecture videos with questions testing perception, comprehension and adaptation of knowledge. Accuracy. Qwen 3.8 Max 88.7%
  • LVBench In use Hours-long video understanding with multiple-choice questions on reasoning, grounding and summarization. Gemini 4 Argon 91.7%
  • MMVU In use 3,000 expert-annotated questions on videos across 27 disciplines. Multiple-choice and open-ended accuracy. Qwen 3.8 Max 82.4%
  • MLVU In use Multi-task long video understanding (minutes to hours) with multiple-choice and generation tasks. Accuracy. Qwen 3.8 Max 90.8%
  • FLEURS In use Models transcribe read speech in 102 languages built from FLoRes-101 sentences. Model cards usually report word error rate, so lower scores are better. Phi-4-multimodal 4.00%
  • VoiceBench In use Spoken instructions with speaker, environment and content variation. Voice assistant responses scored by GPT judge or rules. Qwen3.5-Omni-Plus 93.1%