Best text-to-speech models for a 48GB PC without a graphics card
A leaderboard of the best text-to-speech models for a computer with 48GB of RAM and no graphics card. The best one that runs on it right now is LongCat-AudioDiT-3.5B, followed by LongCat-AudioDiT-1B and OmniVoice. Models are ranked by one score that combines 101 published benchmarks. Models that run on it show the download that fits and the command to run it.
121 models · scores combine 101 benchmarks · updated how it works · find a leaderboard · benchmark your machine
Models for 48GB RAM · no GPU → ~35GB for models (CPU, slow) · reset
+ more
$ terminal
↗ share
detecting your OS…
Score vs API price
x
show
● open weights ◆ API only ─┘ best score for the price · x: USD per 1M characters (log)
| # | model | score↓ | where↓ | speed↓ | sources |
|---|---|---|---|---|---|
| 4 |  Voxtral TTS (4B) | 71±4† | api $16.00/M chars |  | |
| 5 |  OpenAudio S1 | 71±8*† | api $15.00/M chars | | |
| 7 |  Qwen-Audio-3.0-TTS-Plus | 69±2† | api $20.00/M chars | | |
| 8 |  Gemini 3.1 Flash TTS Preview | 69±2 | api $50.00/M chars |  +1 | |
| 11 |  Grok TTS | 68±3 | api $15.00/M chars | | |
| 13 |  Simba 3.2 | 67±3* | api $6.00/M chars | | |
| 15 |  MiniMax Speech 2.8 Turbo | 66±3* | api $60.00/M chars | | |
| 16 |  LongCat-AudioDiT-3.5B · fp32 weights (official LongCat-AudioDiT repo, transformers>=5.3) | 66±2 | local (slow) | | |
| 17 | LongCat-AudioDiT-1B · fp32 weights (official LongCat-AudioDiT repo, transformers>=5.3) | 66±4† | local (slow) | | |
| 18 |  OmniVoice · fp32 | 66±2 | local (slow) | | |
| 19 | MiniMax Speech 2.8 HD | 66±2 | api $100.00/M chars | | |
| 21 |  Eleven v3 | 65±1 | api $100.00/M chars | +2 | |
| 22 |  Chirp 3: HD | 65±4* | api $30.00/M chars | | |
| 23 |  Higgs Audio V3 TTS | 65±3 | api $15.00/M chars | | |
| 25 | Qwen3-TTS 12Hz 1.7B Base · bf16 (qwen-tts) | 65±1 | local (slow) |  +3 | |
| 27 |  VoxCPM2 · bf16 (voxcpm) | 64±1 | local (slow) | | |
| 28 |  Fun-CosyVoice3-0.5B-2512 · fp32 (CosyVoice repo) | 64±1 | local (slow) | | |
| 29 |  Wayu-Paxa-TTS-Edge · fp32 (wayu-tts package) | 64±4† | local (slow) | | |
| 30 | Eleven Turbo v2.5 | 64±4* | api $50.00/M chars | | |
| 31 | Fish Audio S2.1 Pro | 63±3* | api $15.00/M chars | | |
| 32 |  IndexTTS-2 · fp32 (index-tts repo) | 63±1 | local (slow) | +2 | |
| 33 |  MOSS-TTS v1.5 (8B) · fp32 (transformers 5) | 63±2 | local (slow) | | |
| 36 | MiniMax Speech-02-HD | 63±3 | api $100.00/M chars | | |
| 37 | Fish Audio S2 Pro | 63±1 | api $15.00/M chars +local | +3 | |
| 38 | Eleven Flash v2.5 | 63±2 | api $50.00/M chars | | |
| 39 | IndexTTS (v1) · fp32 on CPU (index-tts repo CLI `indextts`) | 62±3 | local (slow) | | |
| 40 |  F5-TTS v1 Base · fp32 | 62±3 | local (slow) | | |
| 41 |  Inworld TTS-1.5 Max | 62±3* | api $10.00/M chars | | |
| 42 | ZipVoice · fp32 (ZipVoice repo) | 62±3 | local (slow) | | |
| 44 | Qwen3-TTS 12Hz 0.6B Base · bf16 (qwen-tts) | 62±2 | local (slow) | | |
| 45 | Eleven Flash v2 | 62±3* | api $50.00/M chars | | |
| 47 | Eleven Turbo v2 | 61±3* | api $50.00/M chars | | |
| 48 | Inworld Realtime TTS-2 | 61±2 | api $25.00/M chars | | |
| 49 | CosyVoice2-0.5B · fp32 (CosyVoice repo) | 61±1 | local (slow) | +1 | |
| 50 | E2 TTS Base (F5-TTS reproduction) · fp32 | 60±2 | local (slow) | | |
| 51 | MiniMax Speech-02-Turbo | 60±3* | api $60.00/M chars | | |
| 52 |  TTS-1 | 59±4 | api $15.00/M chars | | |
| 55 |  CosyVoice v3 Plus | 58±4* | api $26.00/M chars | | |
| 56 |  Octave TTS (Octave 1) | 58±4 | api $50.00/M chars | | |
| 57 |  Sonic 3 | 58±2 | api $65.00/M chars | | |
| 60 | F5-TTS Base (v0) · fp32 | 57±2 | local (slow) | | |
| 61 | TTS-1 HD | 57±7* | api $30.00/M chars | | |
| 62 |  Deepgram Aura 2 | 57±3 | api $30.00/M chars | | |
| 63 |  ZONOS2 8B · q8_0 (zonos2.cpp) | 57±2† | local (slow) | | |
| 64 |  Supertonic 3 · onnx-fp32 (supertonic Python SDK 1.3.x) | 55±10* | local (slow) | | |
| 68 | CosyVoice-300M · fp32 (CosyVoice repo) | 54±3 | local (slow) | | |
| 69 |  Chatterbox Multilingual V3 · fp32 (chatterbox) | 54±10* | local (slow) | | |
| 70 |  VibeVoice 1.5B · bf16 (transformers >=5.17.0) | 53±6 | local (slow) | | |
| 71 |  Lightning v3.1 | 53±3* | api $17.50/M chars | | |
| 73 |  Kokoro 82M v1.0 | 52±3 | api $0.62/M chars +local | | |
| 74 |  LEMAS-TTS (w/ prosody, multilingual_prosody) · official checkpoint (LEMAS-TTS repo, F5-TTS-based) | 52±6*† | local (slow) | | |
| 76 |  Spark-TTS 0.5B · fp32 (Spark-TTS repo CLI) | 52±3 | local (slow) | | |
| 78 |  Gradium TTS | 51±3* | api $38.00/M chars | | |
| 79 |  OpenVoice V1 · fp32 (myshell-ai/OpenVoice) | 51±14* | local (slow) | | |
| 80 | Eleven Multilingual v2 | 51±5 | api $100.00/M chars | | |
| 84 | LEMAS-TTS (w/o prosody, multilingual_grl) · official checkpoint (LEMAS-TTS repo, F5-TTS-based) | 50±6*† | local (slow) | | |
| 85 |  MMS TTS Amharic · fp32 (transformers VitsModel) | 50±16* | local (slow) | | |
| 86 | Chatterbox | 50±5 | api $25.00/M chars +local | | |
| 87 |  Parler-TTS Mini v0.1 · fp32 | 49±14* | local (slow) | | |
| 89 |  SILMA TTS v1 · fp32 checkpoint (silma-tts 1.0.5) | 48±10* | local (slow) | | |
| 92 |  Orpheus TTS 3B 0.1 (finetune-prod) | 47±5 | api $15.00/M chars +local | | |
| 96 |  Piper (en_US lessac medium) · onnx-fp32 (piper-tts 1.8.0) | 47±5* | local (slow) | | |
| 97 |  Vevo (Amphion) · fp32 (Amphion repo) | 46±3 | local (slow) | | |
| 100 |  Amazon Polly Neural | 45±4* | api $16.00/M chars | | |
| 102 | GPT-SoVITS v2 · fp32 (repo api_v2.py) | 44±4 | local (slow) | | |
| 103 | MeloTTS English · fp32 (MeloTTS repo CLI) | 44±5 | local (slow) | | |
| 104 | Parler-TTS Mini v1 · fp32 (parler-tts) | 44±6* | local (slow) | | |
| 106 | Nabra-82M v0.1 · fp32 (Kokoro KModel, Oddadmix/kokoro fork) | 42±10* | local (slow) | | |
| 107 | OpenAudio S1 Mini | 41±5 | api price unknown +local | | |
| 108 | Fish Speech 1.5 · bf16 (fish-speech repo) | 41±3 | local (slow) | | |
| 109 | EMA Lightning · fp32 (PyTorch) | 40±3* | local (slow) | | |
| 113 |  Magpie TTS Multilingual 357M · f16-gguf (NVIDIA NeMo-Speech.cpp) | 39±6* | local (slow) | | |
| 114 | Zonos v0.1 | 39±4 | api price unknown +local | | |
| 115 |  Maya1 · bf16 (transformers + snac) | 38±6* | local (slow) | | |
| 116 | Gemini 3.8 Flash TTS | 37±3* | api price unknown | | |
| 119 | Gemini 3.8 Flash-Lite TTS | 37±3* | api price unknown | | |
| 120 | Eleven v4 | 37±4* | api $80.00/M chars | | |
| 121 | Amphion VALL-E v2 recipe (MLS) · fp32 (Amphion repo) | 36±3 | local (slow) | | |
| 122 |  NeuTTS Air · bf16 (neutts 1.4.1) | 36±10* | local (slow) | | |
| 123 | Eleven v4 Turbo | 36±4* | api $40.00/M chars | | |
| 125 |  Sesame CSM-1B | 34±5 | api $7.00/M chars | | |
| 126 |  VoiceCraft 830M (giga830M) · fp32 (VoiceCraft repo) | 32±4† | local (slow) | | |
| 127 |  VALL-E X (Plachtaa open implementation) · fp32 (Plachtaa/VALL-E-X repo) | 31±4 | local (slow) | | |
| 128 |  OuteTTS 0.2 500M · fp32 (outetts 0.3.3) | 31±6* | local (slow) | | |
| 130 |  StyleTTS 2 (LibriTTS) · fp32 (styletts2 PyPI package 0.1.6) | 30±2 | local (slow) | +2 | |
| 131 |  XTTS v2 · fp32 (coqui-tts 0.27.5) | 30±2 | local (slow) | +7 | |
| 132 | Tortoise TTS v2 · fp32 (coqui-tts 0.27.5) | 29±2 | local (slow) | +2 | |
| 133 | Parler-TTS Large v1 · fp32 | 29±3 | local (slow) | | |
| 137 | Piper (tr_TR dfki medium) · onnx-fp32 (piper-tts) | 26±4* | local (slow) | | |
| 138 | OpenVoice V2 · fp32 (myshell-ai/OpenVoice) | 26±3 | local (slow) | | |
| 139 |  Bark · fp32 | 25±3 | local (slow) | | |
| 140 | SpeechT5 TTS · fp32 (transformers) | 23±3 | local (slow) | | |
| 143 | YourTTS · fp32 (coqui-tts) | 18±3 | local (slow) | | |
| 144 | MMS TTS Turkish · fp32 (transformers VitsModel) | 18±4* | local (slow) | | |
| 147 | Glow-TTS Common Voice (Turkish) · fp32 · CPU | 11±4* | local (slow) | | |
| — | Azure HD 2.5 | no reviewed results | api $22.00/M chars | ||
| — |  Breeze TTS 2 | no reviewed results | api $28.00/M chars | ||
| — | Chatterbox Turbo · fp32 (chatterbox-tts 0.1.7) | no reviewed results | local (slow) | ||
| — |  Dia 1.6B (0626) · fp32 (nari-labs/dia repo) | no reviewed results | local (slow) | ||
| — | Eleven v3 Conversational | no reviewed results | api $40.00/M chars | ||
| — | FastPitch (Coqui LJ Speech) · fp32 · CPU | no reviewed results | local (slow) | ||
| — |  FastSpeech 2 + HiFi-GAN (SpeechBrain LJ Speech) · fp32 · CPU | no reviewed results | local (slow) | ||
| — | Glow-TTS (Coqui LJ Speech) · fp32 · CPU | no reviewed results | local (slow) | ||
| — | Glow-TTS Mai (Ukrainian) · fp32 · CPU | no reviewed results | local (slow) | ||
| — |  Kitten TTS Mini 0.8 · onnx (kittenml 0.9.3) | no reviewed results | local (slow) | ||
| — | KittenTTS 2 · ternary-packed LM (kittenml 0.9.3) | no reviewed results | local (slow) | ||
| — | Lightning v3.1 Pro | no reviewed results | api $19.50/M chars | ||
| — | LightSpeech + Parallel WaveGAN · fp32 · CPU (microsoft/NeuralSpeech LightSpeech repo, LJSpeech test set) | no reviewed results | local (slow) | ||
| — | MiniMax Speech 2.6 HD | no reviewed results | api $100.00/M chars | ||
| — | Neural HMM TTS (Coqui LJ Speech) · fp32 · CPU | no reviewed results | local (slow) | ||
| — | Octave 2 | no reviewed results | api $25.00/M chars | ||
| — | OverFlow (Coqui LJ Speech) · fp32 · CPU | no reviewed results | local (slow) | ||
| — |  Pocket TTS (English) · fp32 (pocket-tts 3.3.0) | no reviewed results | local (slow) | ||
| — |  Polly Generative | no reviewed results | api $30.00/M chars | ||
| — | Sonic 3.6 | no reviewed results | api $38.00/M chars | ||
| — | SpeedySpeech (Coqui LJ Speech) · fp32 · CPU | no reviewed results | local (slow) | ||
| — | Tacotron 2 DDC (Coqui LJ Speech) · fp32 · CPU | no reviewed results | local (slow) | ||
| — | VITS Common Voice (Portuguese) · fp32 · CPU | no reviewed results | local (slow) | ||
| — | VITS CSS10 (French) · fp32 · CPU | no reviewed results | local (slow) | ||
| — | VITS CSS10 (Spanish) · fp32 · CPU | no reviewed results | local (slow) | ||
| — | VITS Thorsten (German) · fp32 · CPU | no reviewed results | local (slow) |
+ 66 more don't fit your machine or budget · show them
187 of 187 catalogue models listed here · sources and licences
* new or thinly tested: ranked on too few benchmark results · † only the maker’s own numbers so far · ~RAM is estimated; full context may need more memory