Skip to content
IFAI

Model rankings

1539 models · 5 boards · 2 sources

‹
Week of September 28, 2026
November 9, 202636 changesNovember 2, 202641 changesOctober 26, 20269 changesOctober 19, 2026247 changesOctober 12, 2026127 changesOctober 5, 202634 changesSeptember 28, 202617 changesSeptember 21, 2026115 changesSeptember 14, 2026174 changesSeptember 7, 2026144 changesAugust 31, 202620 changesAugust 24, 202615 changes
›

Text-to-speech arena · Anonymous head-to-head votes · Updated October 4, 2026

Best voice
Eleven v4ElevenLabs1,321Score
Best open model
Breeze TTS 2BreezeBlue1,216Score
Best from China
Qwen-Audio-3.1-TTS-PlusAlibaba1,292Score
Top three by Score
1Eleven v4ElevenLabs1,321
2Qwen-Audio-3.1-TTS-PlusAlibaba1,292
3Sonic 3.6Cartesia1,278
Top three open models
1Breeze TTS 2BreezeBlue · #10 overall1,216
2Fish Audio S2 ProFish Audio · #32 overall1,117
3Step Audio EditX (Mar 2026)StepFun · #39 overall1,095
This week's changes
Claude Opus 5.5 (Adaptive Reasoning, Max Effort, Default Fallback)Intelligence · #1New
Eleven v4Speech · #1New
Qwen-Audio-3.1-TTS-PlusSpeech · #2New
claude-sonnet-5.5-xhighCoding · #3New
gpt-6.1-sol-maxCoding · #4New
AllOpenClosed
All regionsUSChina
One row per modelAll variants
RankChangeModelScore95% interval
1NewEleven v4ClosedElevenLabs1,321±18
2NewQwen-Audio-3.1-TTS-PlusClosedAlibaba1,292±18
3▼ 2Sonic 3.6ClosedCartesia1,278±16
4▼ 2Gemini 3.8 Flash TTSClosedGoogle1,275±16
5▼ 2Qwen-Audio-3.0-TTS-PlusClosedAlibaba1,261±15
6▼ 2Realtime TTS-2ClosedInworld1,251±16
7▼ 2Simba 3.2ClosedSpeechifyAI1,242±13
8▼ 2Gemini 3.8 Flash-Lite TTSClosedGoogle1,242±15
9▼ 2Luna TTSClosedVUI Labs1,227±13
10▼ 1Breeze TTS 2OpenBreezeBlue1,216±15
11▼ 3Realtime TTS-2 FlashClosedInworld1,214±14
12▼ 1StepAudio 2.5 TTS (Aug 2026)ClosedStepFun1,210±15
13▼ 3Gemini 3.1 Flash TTSClosedGoogle1,209±12
14▼ 2v3 ConversationalClosedElevenLabs1,203±14
15▼ 2Sonic 3.5ClosedCartesia1,190±12
16▼ 1Soniox TTS Real-Time v2ClosedSoniox1,179±11
17—Eleven v3ClosedElevenLabs1,174±11
18▼ 2Speech 2.8 HDClosedMiniMax1,173±11
19▼ 5Lightning V3.1 Pro (Jul 2026)ClosedSmallest.ai1,171±14
20▼ 2Falcon 2ClosedMurf AI1,155±14
21▼ 2Speech 2.8 TurboClosedMiniMax1,153±11
22▼ 2Gradium TTS (Aug 2026)ClosedGradium1,151±14
23▼ 2Fish Audio S2.1 ProClosedFish Audio1,141±13
24▲ 1Async Flash v1.5ClosedAsync1,136±13
25▼ 1SpaceXAI TTSClosedSpaceXAI1,136±15
26▼ 4Step TTS 2 (Mar 2026)ClosedStepFun1,134±12
27▼ 4Azure HD 2.5ClosedMicrosoft1,131±12
28▼ 2Speech 2.6 HDClosedMiniMax1,128±11
29▼ 1Simba 3.0ClosedSpeechifyAI1,125±12
30▼ 1Speech 2.6 TurboClosedMiniMax1,124±11
31▼ 1Async Pro v1.0ClosedAsync1,119±13
32▼ 5Fish Audio S2 ProOpenFish Audio1,117±12
33▼ 2Speech-02-HDClosedMiniMax1,114±11
34—Speech 2.6ClosedHithink1,103±13
35▼ 3TTS-1 HDClosedOpenAI1,103±12
36▼ 3Chatterbox HDClosedResemble AI1,102±13
37▼ 2Turbo v2.5ClosedElevenLabs1,098±11
38▼ 1Multilingual v2ClosedElevenLabs1,097±10
39▼ 3Step Audio EditX (Mar 2026)OpenStepFun1,095±13
40▼ 2TTS-1ClosedOpenAI1,090±11
41▼ 2Speech-02-TurboClosedMiniMax1,090±12
42—Voxtral TTSOpenMistral1,083±13
43▼ 3Gemini 2.5 Flash Lite TTSClosedGoogle1,083±12
44▼ 3OpenAudio S1ClosedFish Audio1,082±11
45▼ 2StudioClosedGoogle1,079±11
46▼ 2Flash v2.5ClosedElevenLabs1,078±11
47▼ 1GPT-Realtime-2ClosedOpenAI1,077±13
48▼ 3JourneyClosedGoogle1,074±13
49▼ 2Maya 2 FlashClosedMaya Research1,071±14
50—T2A-01-HDClosedMiniMax1,066±11

Intelligence over time

Each maker's best model at the time on the Artificial Analysis Intelligence Index, by the month it was released.

Gained most · Anthropic +57.2Highest now · Anthropic 64.6Gap between highest and lowest 18.1 ptsTime range Nov 2025 – Nov 2026
64.6Anthropic60.0Google56.9OpenAI50.1Meta46.5Kimi

Arena data comes from the leaderboard dataset it publishes on Hugging Face (CC BY 4.0). Only each board's overall leaderboard is used; scores and ranks are unchanged. dataset · CC BY 4.0

Artificial Analysis data comes from its API (artificialanalysis.ai). Scores are unchanged; ranks follow the scores. artificialanalysis.ai

Open or closed follows the license Arena gives; prices and Artificial Analysis intelligence show only where model names match, otherwise “—”.