Coming soonHosted open-source TTS

Every open-source TTS,
one API.

All 22 models on openspeech.dev, behind a single endpoint. No CUDA, no Docker, no deploys. Pay only for what you generate.

Pricing

Pay as you go. No subscription.

Add a balance, generate audio, pay per minute. Three model tiers, three prices. That’s the whole pricing page.

Small models

$0.10/min

Lightweight, fast, low-latency.

KokoroPiperMeloTTSKittenTTSStyleTTS 2

Medium models

$0.20/min

The popular mid-tier. Voice cloning, multilingual.

ChatterboxOrpheusXTTS v2F5-TTSParler-TTSOpenVoice v2Spark-TTSMegaTTS 3

Large models

$0.40/min

Heavy hitters. Long-form, dialogue, expressive.

BarkDiaFish SpeechVibeVoiceCosyVoice 2IndexTTS 2Higgs Audio v2Qwen3-TTSWhisperSpeech

$10

deposit

≈ 100 min small

≈ 50 min medium

≈ 25 min large

$25

deposit

≈ 250 min small

≈ 125 min medium

≈ 62.5 min large

$50

deposit

≈ 500 min small

≈ 250 min medium

≈ 125 min large

Free tier: 5 minutes per day across small + medium models. No card required. Balance never expires. Cancel anytime — there’s nothing to cancel.

Try any model in seconds

The OSS directory shows you which model exists. We make running one not require Docker, CUDA, or a GPU.

Cheaper than ElevenLabs

Small-model tier is 2-3× cheaper. Large-model tier still undercuts ElevenLabs Creator. Pay only for what you generate.

Open underneath

The directory at openspeech.dev stays open-source and community-maintained. The hosting business pays for keeping it running.