ElevenLabsAudio matchupSpeech output
Eleven v3vs.Eleven v4
Which model is better?
Eleven v4 does better in independent tests, while Eleven v3 is more popular.
Pick Eleven v3 for
Pick Eleven v4 for
Shared tests
Independent evaluators and leaderboards only, the newest shared snapshot per evaluator. With 95 % ranges a lead counts when they do not overlap; for hit rates, when it holds in a 95 % test. Without any uncertainty figure at most “slightly ahead”.
Askable Labs VOICE-H
Eleven v4 aheadEleven v4 is clearly ahead in 12 of 12 subtests, Eleven v3 in none.
Askable Labs VOICE-H naturalness Elo (Spanish)
95 % ranges 903 to 960 and 1,203 to 1,260, as of October 17, 2026
Askable Labs VOICE-H emotion/tone Elo (Spanish)
95 % ranges 916 to 974 and 1,213 to 1,273, as of October 17, 2026
Askable Labs VOICE-H overall preference Elo (Spanish)
95 % ranges 931 to 975 and 1,159 to 1,206, as of October 17, 2026
Show all 12 subtests
| Subtest | Eleven v3 | Eleven v4 | Verdict |
|---|---|---|---|
| Askable Labs VOICE-H naturalness Elo (Spanish) | 933 | 1,231 | Eleven v4 ahead |
| Askable Labs VOICE-H emotion/tone Elo (Spanish) | 946 | 1,243 | Eleven v4 ahead |
| Askable Labs VOICE-H overall preference Elo (Spanish) | 953 | 1,182 | Eleven v4 ahead |
| Askable Labs VOICE-H pronunciation accuracy Elo (Spanish) | 1,015 | 1,231 | Eleven v4 ahead |
| Askable Labs VOICE-H emotion/tone Elo (French) | 1,133 | 1,309 | Eleven v4 ahead |
| Askable Labs VOICE-H naturalness Elo (English) | 980 | 1,116 | Eleven v4 ahead |
| Askable Labs VOICE-H emotion/tone Elo (English) | 1,006 | 1,144 | Eleven v4 ahead |
| Askable Labs VOICE-H naturalness Elo (French) | 1,146 | 1,284 | Eleven v4 ahead |
| Askable Labs VOICE-H pronunciation accuracy Elo (English) | 998 | 1,117 | Eleven v4 ahead |
| Askable Labs VOICE-H overall preference Elo (English) | 994 | 1,109 | Eleven v4 ahead |
| Askable Labs VOICE-H pronunciation accuracy Elo (French) | 1,153 | 1,287 | Eleven v4 ahead |
| Askable Labs VOICE-H overall preference Elo (French) | 1,138 | 1,235 | Eleven v4 ahead |
Source · latest October 17, 2026
VoiceArena
Eleven v4 aheadEleven v4 is clearly ahead in 7 of 7 subtests, Eleven v3 in none.
VoiceArena TTS Elo: Arabic MSA
95 % ranges 997 to 1,017 and 1,110 to 1,154, as of October 28, 2026
VoiceArena TTS Elo: Japanese
95 % ranges 1,021 to 1,041 and 1,135 to 1,173, as of October 28, 2026
VoiceArena TTS Elo: US English
95 % ranges 977 to 991 and 1,053 to 1,079, as of October 28, 2026
Show all 7 subtests
| Subtest | Eleven v3 | Eleven v4 | Verdict |
|---|---|---|---|
| VoiceArena TTS Elo: Arabic MSA | 1,007 | 1,132 | Eleven v4 ahead |
| VoiceArena TTS Elo: Japanese | 1,031 | 1,154 | Eleven v4 ahead |
| VoiceArena TTS Elo: US English | 984 | 1,066 | Eleven v4 ahead |
| VoiceArena TTS Elo: Hindi | 1,044 | 1,114 | Eleven v4 ahead |
| VoiceArena TTS Elo: Mexican Spanish | 1,007 | 1,068 | Eleven v4 ahead |
| VoiceArena TTS Elo: Vietnamese | 1,033 | 1,083 | Eleven v4 ahead |
| VoiceArena TTS Elo: Brazilian Portuguese | 1,034 | 1,070 | Eleven v4 ahead |
Source · latest October 28, 2026
Artificial Analysis
Eleven v4 aheadEleven v4 is clearly ahead in 2 of 2 subtests, Eleven v3 in none.
Artificial Analysis Provider Voice Arena preference
95 % ranges 1,177 to 1,197 and 1,319 to 1,349, as of November 3, 2026
Artificial Analysis Controlled Voice Arena Elo
95 % ranges 1,063 to 1,085 and 1,143 to 1,167, as of November 3, 2026
Source · latest November 3, 2026
DesignArena
Eleven v4 slightly aheadEleven v4 is clearly ahead in 0 of 2 subtests and slightly in 2, Eleven v3 in none.
DesignArena TTS Arena Elo
as of October 28, 2026
DesignArena Audio Realism Elo
as of October 28, 2026
Source · latest October 28, 2026
All data side by side
Prices are verbatim from the provider, because audio models bill per minute, character or token.