Text to speech

Speech-synthesis models ranked by blind listening arenas, voice apps for narration and voice cloning, and open models with local servers and UIs.

Models are ranked by Artificial Analysis Speech Arena: Text to Speech (provider voices), then TTS Arena V2 (Hugging Face Space) for models without a result on the first. Tools inherit the tier of the best model they offer. How we rank.

Best overallSEleven v4 Turbo1,329 on Artificial Analysis Speech Arena: Text to Speech (provider voices)
Best valueASimba 3.2Tier A for $6.58/M characters
Best free startSElevenLabsFree tier · from $6/mo · via Eleven v4 Turbo
Best local (≤24 GB GPU)BBreeze TTS 2~12 GB VRAM · 1,222 on Artificial Analysis Speech Arena: Text to Speech (provider voices)

Tier: how close to the best result on the task's main public benchmark
Outlined: a tool inherits the tier of the best model it offers
No public result yet
Cloud · Local: runs on your own hardware

Models

The engines themselves. Use them through an API or inside the apps on the right.

Cloud

Eleven v4 TurboElevenLabs · $40/M characters1,329 on Artificial Analysis Speech Arena: Text to Speech (provider voices)$S

Eleven v4ElevenLabs · $80/M characters1,328 on Artificial Analysis Speech Arena: Text to Speech (provider voices)S

Qwen-Audio-3.1-TTS-PlusAlibaba (Qwen) · $19/M characters1,296 on Artificial Analysis Speech Arena: Text to Speech (provider voices)$A

Sonic 3.6Cartesia · $49/M characters1,287 on Artificial Analysis Speech Arena: Text to Speech (provider voices)A

Gemini 3.8 Flash TTSGoogle · $16/M characters1,278 on Artificial Analysis Speech Arena: Text to Speech (provider voices)$A

Realtime TTS-2Inworld AI · $21/M characters1,262 on Artificial Analysis Speech Arena: Text to Speech (provider voices)A

Gemini 3.8 Flash-Lite TTSGoogle · $11/M characters1,248 on Artificial Analysis Speech Arena: Text to Speech (provider voices)$A

Simba 3.2Speechify · $6.58/M characters1,245 on Artificial Analysis Speech Arena: Text to Speech (provider voices)$B

Luna TTSVUI Labs · $15/M characters1,217 on Artificial Analysis Speech Arena: Text to Speech (provider voices)B

StepAudio 2.5 TTS (Aug 2026)StepFun · $85/M characters1,216 on Artificial Analysis Speech Arena: Text to Speech (provider voices)B

Fish Audio S2.1 ProFish Audio · $15/M characters1,190 on Artificial Analysis Speech Arena: Text to Speech (provider voices)B

Soniox TTS Real-Time v2Soniox · $14/M characters1,179 on Artificial Analysis Speech Arena: Text to Speech (provider voices)B

Gradium TTS (Sep 2026)Gradium · $48/M characters1,174 on Artificial Analysis Speech Arena: Text to Speech (provider voices)B

Speech 2.8 HDMiniMax · $100/M characters1,174 on Artificial Analysis Speech Arena: Text to Speech (provider voices)B

Lightning V3.1 Pro (Jul 2026)Smallest.ai · $20/M characters1,170 on Artificial Analysis Speech Arena: Text to Speech (provider voices)B

Falcon 2Murf AI · $10/M characters1,154 on Artificial Analysis Speech Arena: Text to Speech (provider voices)C

SpaceXAI TTS (AA name)xAI · $15/M characters1,146 on Artificial Analysis Speech Arena: Text to Speech (provider voices)C

Async Flash v1.5Async · $10/M characters1,141 on Artificial Analysis Speech Arena: Text to Speech (provider voices)C

Azure HD 2.5Microsoft · $22/M characters1,137 on Artificial Analysis Speech Arena: Text to Speech (provider voices)C

Async Pro v1.0Async · $20/M characters1,129 on Artificial Analysis Speech Arena: Text to Speech (provider voices)C

Maya 2Maya Research1,065 on Artificial Analysis Speech Arena: Text to Speech (provider voices)C

Polly GenerativeAmazon · $30/M characters1,064 on Artificial Analysis Speech Arena: Text to Speech (provider voices)C

CodaRime · $50/M characters1,057 on Artificial Analysis Speech Arena: Text to Speech (provider voices)C

Octave 2 (preview)Hume AI · $88/M characters1,055 on Artificial Analysis Speech Arena: Text to Speech (provider voices)C

MiMo-V2.5-TTSXiaomi1,054 on Artificial Analysis Speech Arena: Text to Speech (provider voices)C

Flux TTSDeepgram · $45/M characters1,052 on Artificial Analysis Speech Arena: Text to Speech (provider voices)C

Bland Speech v3Bland AI · $40/M characters1,039 on Artificial Analysis Speech Arena: Text to Speech (provider voices)C

MAI-Voice-1Microsoft AI · $22/M characters1,031 on Artificial Analysis Speech Arena: Text to Speech (provider voices)C

Kugel 3KugelAudio931 on Artificial Analysis Speech Arena: Text to Speech (provider voices)–

GPT-4o Mini TTS (2025-12-15)OpenAINo public result yet –

Mist v3RimeNo public result yet –

SSFM 3.0 (ssfm-v30)Neosapience (Typecast)No public result yet

Local

Fish Audio Drama 3 (Preview)Fish Audio · ~48 GB VRAM1,244 on Artificial Analysis Speech Arena: Text to Speech (provider voices)$B

Breeze TTS 2BreezeBlue · ~12 GB VRAM1,222 on Artificial Analysis Speech Arena: Text to Speech (provider voices)$C

Step-Audio-EditXStepFun · ~8 GB VRAM1,100 on Artificial Analysis Speech Arena: Text to Speech (provider voices)$C

Voxtral TTS (4B, 2603)Mistral AI · ~10 GB VRAM1,089 on Artificial Analysis Speech Arena: Text to Speech (provider voices)C

Kokoro-82M v1.0hexgrad · ~1 GB VRAM1,064 on Artificial Analysis Speech Arena: Text to Speech (provider voices)$C

Magpie-Multilingual 357M (Feb 2026)NVIDIA · ~1 GB VRAM1,064 on Artificial Analysis Speech Arena: Text to Speech (provider voices)C

Maya1Maya Research · ~8 GB VRAM1,046 on Artificial Analysis Speech Arena: Text to Speech (provider voices)C

Higgs TTS 3 (4B)Boson AI · ~12 GB VRAM1,042 on Artificial Analysis Speech Arena: Text to Speech (provider voices)C

VibeVoice-1.5BMicrosoft · ~8 GB VRAM957 on Artificial Analysis Speech Arena: Text to Speech (provider voices)–

Chatterbox Multilingual V3Resemble AI · ~2 GB VRAMNo public result yet –

Dia2-2BNari Labs · ~6 GB VRAMNo public result yet –

F5-TTS v1SWivid (Shanghai Jiao Tong University) · Open weightsNo public result yet –

Fun-CosyVoice3-0.5B-2512Alibaba (FunAudioLLM) · ~2 GB VRAMNo public result yet –

IndexTTS-2.5bilibili (IndexTeam) · Open weightsNo public result yet –

Kitten TTS 2KittenML · Open weightsNo public result yet –

MOSS-TTS v1.5OpenMOSS (Fudan University / MOSI) · ~19 GB VRAMNo public result yet –

NeuTTS AirNeuphonic · ~2 GB VRAMNo public result yet –

OmniVoicek2-fsa (Next-gen Kaldi) · ~2 GB VRAMNo public result yet –

Orpheus 3B 0.1Canopy Labs · ~9 GB VRAMNo public result yet –

Pocket TTSKyutai · Open weightsNo public result yet –

Qwen3-TTS-12Hz-1.7BAlibaba (Qwen) · ~6 GB VRAMNo public result yet –

VoxCPM2OpenBMB · ~6 GB VRAMNo public result yet –

ZONOS2Zyphra · Open weightsNo public result yet

Apps

Ready-made apps with a UI: sign in or install, then start.

Cloud

ElevenLabsElevenLabs · Free tier · $6/movia Eleven v4 Turbo$A

SpeechifySpeechify · Free tier · $29/movia Simba 3.2 A

Fish AudioFish Audio · Free tier · $15/movia Fish Audio Drama 3 (Preview)B

MurfMurf AI · Free tier · $29/movia Falcon 2 –

DescriptDescript · Free tier · $24/moNo public result yet –

NarakeetNarakeetNo public result yet –

NaturalReaderNaturalSoft · Free tierNo public result yet –

TypecastNeosapience · Free tier · $5/moNo public result yet –

WellSaidWellSaid Labs · Free tier · $19/moNo public result yet

Local

VoiceboxJamie Pine (open source) · Freevia Kokoro-82M v1.0$–

ebook2audiobookDrewThomasson (open source) · FreeNo public result yet

Local servers & APIs

Run open voice models on your own machine, or call speech APIs from your own apps.

Cloud

Nothing tracked yet

Local

C sherpa-onnxk2-fsa (Next-gen Kaldi) · Freevia Kokoro-82M v1.0$C

Kokoro-FastAPIremsky (open source) · Freevia Kokoro-82M v1.0$C

Speachesspeaches-ai (open source) · Freevia Kokoro-82M v1.0$C

TTS WebUIrsxdalv (open source) · Freevia Kokoro-82M v1.0$C

NVIDIA NeMo SpeechNVIDIA · Freevia Magpie-Multilingual 357M (Feb 2026)–

PiperOpen Home Foundation · FreeNo public result yet –

AllTalk TTSerew123 (open source) · FreeNo public result yet

Quality vs. price

Artificial Analysis Speech Arena: Text to Speech (provider voices) score against API price (USD per 1M characters, log scale). Up and to the left is better; the line joins the models that nothing else beats on both.

Closed models
Open-weights models
Best-value frontier

Show as table

Model	Weights	Artificial Analysis Speech Arena: Text to Speech (provider voices)	Price	Frontier
Eleven v4 Turbo	Closed	1,329	$40/M characters	Yes
Eleven v4	Closed	1,328	$80/M characters
Qwen-Audio-3.1-TTS-Plus	Closed	1,296	$19/M characters	Yes
Sonic 3.6	Closed	1,287	$49/M characters
Gemini 3.8 Flash TTS	Closed	1,278	$16/M characters	Yes
Realtime TTS-2	Closed	1,262	$21/M characters
Gemini 3.8 Flash-Lite TTS	Closed	1,248	$11/M characters	Yes
Simba 3.2	Closed	1,245	$6.58/M characters	Yes
Fish Audio Drama 3 (Preview)	Open	1,244	$15/M characters
Breeze TTS 2	Open	1,222	$34/M characters
Luna TTS	Closed	1,217	$15/M characters
StepAudio 2.5 TTS (Aug 2026)	Closed	1,216	$85/M characters
Fish Audio S2.1 Pro	Closed	1,190	$15/M characters
Soniox TTS Real-Time v2	Closed	1,179	$14/M characters
Gradium TTS (Sep 2026)	Closed	1,174	$48/M characters
Speech 2.8 HD	Closed	1,174	$100/M characters
Lightning V3.1 Pro (Jul 2026)	Closed	1,170	$20/M characters
Falcon 2	Closed	1,154	$10/M characters
SpaceXAI TTS (AA name)	Closed	1,146	$15/M characters
Async Flash v1.5	Closed	1,141	$10/M characters
Azure HD 2.5	Closed	1,137	$22/M characters
Async Pro v1.0	Closed	1,129	$20/M characters
Voxtral TTS (4B, 2603)	Open	1,089	$16/M characters
Kokoro-82M v1.0	Open	1,064	$0.65/M characters	Yes
Polly Generative	Closed	1,064	$30/M characters
Coda	Closed	1,057	$50/M characters
Octave 2 (preview)	Closed	1,055	$88/M characters
Flux TTS	Closed	1,052	$45/M characters
Higgs TTS 3 (4B)	Open	1,042	$15/M characters
Bland Speech v3	Closed	1,039	$40/M characters
MAI-Voice-1	Closed	1,031	$22/M characters

Quality vs. setup

The same tools as the map, placed by how good the results can be (up) and how much setup they take (right). Top-left is the easiest way to top quality; further right buys control, privacy or price at the cost of setup. Models are left out: you reach them through these tools.

Cloud
Local: runs on your hardware
Hollow: tier inherited from the best model the tool offers

Not placed (no public result yet): ebook2audiobook, Piper, AllTalk TTS, Descript, Narakeet, NaturalReader, Typecast, WellSaid

Everything, in order

All text to speech options in quality order
Tier	Name	Approach	Runs	Rated by	Price	Skill
S	Eleven v4 Turbo ElevenLabs	Models	Cloud	1,329 on Artificial Analysis Speech Arena: Text to Speech (provider voices)	$40/M characters	—
S	Eleven v4 ElevenLabs	Models	Cloud	1,328 on Artificial Analysis Speech Arena: Text to Speech (provider voices)	$80/M characters	—
S	Qwen-Audio-3.1-TTS-Plus Alibaba (Qwen)	Models	Cloud	1,296 on Artificial Analysis Speech Arena: Text to Speech (provider voices)	$19/M characters	—
S	ElevenLabs ElevenLabs	Apps	Cloud	via Eleven v4 Turbo	Free tier · $6/mo	beginner
A	Sonic 3.6 Cartesia	Models	Cloud	1,287 on Artificial Analysis Speech Arena: Text to Speech (provider voices)	$49/M characters	—
A	Gemini 3.8 Flash TTS Google	Models	Cloud	1,278 on Artificial Analysis Speech Arena: Text to Speech (provider voices)	$16/M characters	—
A	Realtime TTS-2 Inworld AI	Models	Cloud	1,262 on Artificial Analysis Speech Arena: Text to Speech (provider voices)	$21/M characters	—
A	Gemini 3.8 Flash-Lite TTS Google	Models	Cloud	1,248 on Artificial Analysis Speech Arena: Text to Speech (provider voices)	$11/M characters	—
A	Simba 3.2 Speechify	Models	Cloud	1,245 on Artificial Analysis Speech Arena: Text to Speech (provider voices)	$6.58/M characters	—
A	Fish Audio Drama 3 (Preview) Fish Audio	Models	Local	1,244 on Artificial Analysis Speech Arena: Text to Speech (provider voices)	~48 GB VRAM	—
A	Speechify Speechify	Apps	Cloud	via Simba 3.2	Free tier · $29/mo	beginner
A	Fish Audio Fish Audio	Apps	Cloud	via Fish Audio Drama 3 (Preview)	Free tier · $15/mo	beginner
B	Breeze TTS 2 BreezeBlue	Models	Local	1,222 on Artificial Analysis Speech Arena: Text to Speech (provider voices)	~12 GB VRAM	—
B	Luna TTS VUI Labs	Models	Cloud	1,217 on Artificial Analysis Speech Arena: Text to Speech (provider voices)	$15/M characters	—
B	StepAudio 2.5 TTS (Aug 2026) StepFun	Models	Cloud	1,216 on Artificial Analysis Speech Arena: Text to Speech (provider voices)	$85/M characters	—
B	Fish Audio S2.1 Pro Fish Audio	Models	Cloud	1,190 on Artificial Analysis Speech Arena: Text to Speech (provider voices)	$15/M characters	—
B	Soniox TTS Real-Time v2 Soniox	Models	Cloud	1,179 on Artificial Analysis Speech Arena: Text to Speech (provider voices)	$14/M characters	—
B	Gradium TTS (Sep 2026) Gradium	Models	Cloud	1,174 on Artificial Analysis Speech Arena: Text to Speech (provider voices)	$48/M characters	—
B	Speech 2.8 HD MiniMax	Models	Cloud	1,174 on Artificial Analysis Speech Arena: Text to Speech (provider voices)	$100/M characters	—
B	Lightning V3.1 Pro (Jul 2026) Smallest.ai	Models	Cloud	1,170 on Artificial Analysis Speech Arena: Text to Speech (provider voices)	$20/M characters	—
B	Falcon 2 Murf AI	Models	Cloud	1,154 on Artificial Analysis Speech Arena: Text to Speech (provider voices)	$10/M characters	—
B	Murf Murf AI	Apps	Cloud	via Falcon 2	Free tier · $29/mo	beginner
C	SpaceXAI TTS (AA name) xAI	Models	Cloud	1,146 on Artificial Analysis Speech Arena: Text to Speech (provider voices)	$15/M characters	—
C	Async Flash v1.5 Async	Models	Cloud	1,141 on Artificial Analysis Speech Arena: Text to Speech (provider voices)	$10/M characters	—
C	Azure HD 2.5 Microsoft	Models	Cloud	1,137 on Artificial Analysis Speech Arena: Text to Speech (provider voices)	$22/M characters	—
C	Async Pro v1.0 Async	Models	Cloud	1,129 on Artificial Analysis Speech Arena: Text to Speech (provider voices)	$20/M characters	—
C	Step-Audio-EditX StepFun	Models	Local	1,100 on Artificial Analysis Speech Arena: Text to Speech (provider voices)	~8 GB VRAM	—
C	Voxtral TTS (4B, 2603) Mistral AI	Models	Local	1,089 on Artificial Analysis Speech Arena: Text to Speech (provider voices)	~10 GB VRAM	—
C	Maya 2 Maya Research	Models	Cloud	1,065 on Artificial Analysis Speech Arena: Text to Speech (provider voices)	—	—
C	Kokoro-82M v1.0 hexgrad	Models	Local	1,064 on Artificial Analysis Speech Arena: Text to Speech (provider voices)	~1 GB VRAM	—
C	Polly Generative Amazon	Models	Cloud	1,064 on Artificial Analysis Speech Arena: Text to Speech (provider voices)	$30/M characters	—
C	Magpie-Multilingual 357M (Feb 2026) NVIDIA	Models	Local	1,064 on Artificial Analysis Speech Arena: Text to Speech (provider voices)	~1 GB VRAM	—
C	Coda Rime	Models	Cloud	1,057 on Artificial Analysis Speech Arena: Text to Speech (provider voices)	$50/M characters	—
C	Octave 2 (preview) Hume AI	Models	Cloud	1,055 on Artificial Analysis Speech Arena: Text to Speech (provider voices)	$88/M characters	—
C	MiMo-V2.5-TTS Xiaomi	Models	Cloud	1,054 on Artificial Analysis Speech Arena: Text to Speech (provider voices)	—	—
C	Flux TTS Deepgram	Models	Cloud	1,052 on Artificial Analysis Speech Arena: Text to Speech (provider voices)	$45/M characters	—
C	Maya1 Maya Research	Models	Local	1,046 on Artificial Analysis Speech Arena: Text to Speech (provider voices)	~8 GB VRAM	—
C	Higgs TTS 3 (4B) Boson AI	Models	Local	1,042 on Artificial Analysis Speech Arena: Text to Speech (provider voices)	~12 GB VRAM	—
C	Bland Speech v3 Bland AI	Models	Cloud	1,039 on Artificial Analysis Speech Arena: Text to Speech (provider voices)	$40/M characters	—
C	MAI-Voice-1 Microsoft AI	Models	Cloud	1,031 on Artificial Analysis Speech Arena: Text to Speech (provider voices)	$22/M characters	—
C	VibeVoice-1.5B Microsoft	Models	Local	957 on Artificial Analysis Speech Arena: Text to Speech (provider voices)	~8 GB VRAM	—
C	Kugel 3 KugelAudio	Models	Cloud	931 on Artificial Analysis Speech Arena: Text to Speech (provider voices)	—	—
C	Voicebox Jamie Pine (open source)	Apps	Local	via Kokoro-82M v1.0	Free	beginner
C	sherpa-onnx k2-fsa (Next-gen Kaldi)	Local servers & APIs	Local	via Kokoro-82M v1.0	Free	advanced
C	Kokoro-FastAPI remsky (open source)	Local servers & APIs	Local	via Kokoro-82M v1.0	Free	intermediate
C	Speaches speaches-ai (open source)	Local servers & APIs	Local	via Kokoro-82M v1.0	Free	advanced
C	TTS WebUI rsxdalv (open source)	Local servers & APIs	Local	via Kokoro-82M v1.0	Free	intermediate
C	NVIDIA NeMo Speech NVIDIA	Local servers & APIs	Local	via Magpie-Multilingual 357M (Feb 2026)	Free	advanced
–	ebook2audiobook DrewThomasson (open source)	Apps	Local	No public result yet	Free	intermediate
–	Piper Open Home Foundation	Local servers & APIs	Local	No public result yet	Free	intermediate
–	AllTalk TTS erew123 (open source)	Local servers & APIs	Local	No public result yet	Free	advanced
–	Chatterbox Multilingual V3 Resemble AI	Models	Local	No public result yet	~2 GB VRAM	—
–	Descript Descript	Apps	Cloud	No public result yet	Free tier · $24/mo	beginner
–	Dia2-2B Nari Labs	Models	Local	No public result yet	~6 GB VRAM	—
–	F5-TTS v1 SWivid (Shanghai Jiao Tong University)	Models	Local	No public result yet	Open weights	—
–	Fun-CosyVoice3-0.5B-2512 Alibaba (FunAudioLLM)	Models	Local	No public result yet	~2 GB VRAM	—
–	GPT-4o Mini TTS (2025-12-15) OpenAI	Models	Cloud	No public result yet	—	—
–	IndexTTS-2.5 bilibili (IndexTeam)	Models	Local	No public result yet	Open weights	—
–	Kitten TTS 2 KittenML	Models	Local	No public result yet	Open weights	—
–	Mist v3 Rime	Models	Cloud	No public result yet	—	—
–	MOSS-TTS v1.5 OpenMOSS (Fudan University / MOSI)	Models	Local	No public result yet	~19 GB VRAM	—
–	Narakeet Narakeet	Apps	Cloud	No public result yet	—	beginner
–	NaturalReader NaturalSoft	Apps	Cloud	No public result yet	Free tier	beginner
–	NeuTTS Air Neuphonic	Models	Local	No public result yet	~2 GB VRAM	—
–	OmniVoice k2-fsa (Next-gen Kaldi)	Models	Local	No public result yet	~2 GB VRAM	—
–	Orpheus 3B 0.1 Canopy Labs	Models	Local	No public result yet	~9 GB VRAM	—
–	Pocket TTS Kyutai	Models	Local	No public result yet	Open weights	—
–	Qwen3-TTS-12Hz-1.7B Alibaba (Qwen)	Models	Local	No public result yet	~6 GB VRAM	—
–	SSFM 3.0 (ssfm-v30) Neosapience (Typecast)	Models	Cloud	No public result yet	—	—
–	Typecast Neosapience	Apps	Cloud	No public result yet	Free tier · $5/mo	beginner
–	VoxCPM2 OpenBMB	Models	Local	No public result yet	~6 GB VRAM	—
–	WellSaid WellSaid Labs	Apps	Cloud	No public result yet	Free tier · $19/mo	beginner
–	ZONOS2 Zyphra	Models	Local	No public result yet	Open weights	—

Benchmarks used

Artificial Analysis Speech Arena: Text to Speech (provider voices) by Artificial Analysis (blind human votes, Elo from blind pairwise listener votes, each model using its own provider voices (all categories, US and UK accents)) · snapshot 19 Oct 2026
96 ranked models read from the page (Provider Voice Leaderboard, all categories and accents). Hume's Octave 2 (rank 61, 1055) and Octave TTS (rank 69, 1039) are still listed although Hume is sunsetting its TTS API on 2026-11-13. Missing: GPT-4o Mini TTS, Chatterbox Multilingual V3, Qwen3-TTS open weights, Dia2. Leader value shown rounded (1329) on 2026-10-19; exact value carried over from the 2026-10-18 read. No last-updated date shown, so as_of = retrieval date.
TTS Arena V2 (Hugging Face Space) by TTS-AGI (TTS Arena) (blind human votes, Bradley-Terry rating (Elo-like, centred on 1500) from blind pairwise votes on random prompts; the board sorts by the lower bound of the confidence interval) · snapshot 18 Oct 2026
38 ranked entries (plus 1 suspended) read from the Space's public JSON endpoint https://tts-agi-tts-arena-v2.hf.space/api/leaderboard. Many entries are older versions (Eleven v3, Multilingual v2, Cartesia Sonic 2, Inworld TTS 1.5) and several are retired; the rank-1 entry (lower-bound sort) is not the highest raw rating. No snapshot date shown, so as_of = retrieval date.