ElevenLabs alternative: open Text to speech models on your own GPU
ElevenLabs is a hosted service for Text to speech. If you would rather not pay per generation or per month, you can run open models that do the same job on a GPU you rent by the second.
Open models you can use instead
Kokoro-82M80M · 2 GBaudio.cpp-gguf20M · 2 GBOmniVoice610M · 3 GBVieNeu-TTS-v3-Turbo130M · 2 GBkokoro-inno-clone-tuner10M · 2 GBQwen3-TTS-GGUF910M · 2 GBQwen3-TTS-12Hz-1.7B-CustomVoice1.9B · 6 GBHiggs-Audio-v3-Studio4.7B · 5 GB
All Text to speech modelsGPUs that fit these models today
| GPU | Memory | Per hour |
|---|---|---|
| GTX 1660 S | 6 GB | $0.11 |
| GTX 1650 | 4 GB | $0.13 |
| GTX 1070 | 8 GB | $0.14 |
| GTX 1660 | 6 GB | $0.14 |
| RTX A2000 | 6 GB | $0.15 |
Live starting prices from our catalog. You pay per second while the machine runs.
Why switch
- No per-generation credits or monthly plan: you pay only while the GPU runs.
- Choose the model, settings and resolution yourself, with no queue.
- Your prompts and files stay on your own machine.
Questions
- Is there an open alternative to ElevenLabs?
- Yes. Open Text to speech models are free to use under their licenses; you only pay for the GPU time while they run, from $0.11 per hour.
- Do I need a subscription?
- No. You pay only for the time your machine runs, from your USD balance. Stop it whenever you are done.
Try it on your own GPU in minutes
Pick a GPU, open Jupyter or SSH and run any open model. You pay per second and can stop anytime; no subscription, no setup fee.
- Per-second billing in USD
- PyTorch and CUDA ready in minutes
- Rent from ChatGPT or Claude too
ElevenLabs is a trademark of its owner. CoreHu is not affiliated with or endorsed by ElevenLabs.