Replicate API Pricing

Short answer

Replicate hosts open and commercial TTS models with runtime-based job pricing.

Model Input /1M Output /1M Cached in /1M Context
— — — ~75ms model latency; 32 languages
Speech 2.8 Turbo Replicate
— — — 40+ languages; streaming; voice cloning
Speech 2.8 HD Replicate
— — — Studio-quality; sound tags; voice cloning
Realtime TTS-2 Replicate
— — — 100+ languages; steering; instant cloning
Chatterbox Replicate
— — — Standard, Multilingual, and Turbo; voice cloning
Qwen3-TTS Replicate
— — — Flash, realtime, instruct, voice clone/design variants
HunyuanVideo 1.5 Replicate
— — — up to 1080p; 5/8/10s; 8.3B (open weights)