Cartesia
Listed at https://cartesia.ai
Overall Rank #28 ⭐ Consider
⚠️ Partially available (stable proxy recommended in China) | 🌍 International
💰 Token Pricing
| Type | Price | Note |
|---|---|---|
| Input | Sonic: $0.03/1K chars (Standard), $0.06/1K chars (Pro), $0.12/1K chars (Turbo); Sonic 2: $0.04-0.15/1K chars | per million tokens |
| Output | TTS character-based pricing (output = audio), per-second for streaming | per million tokens |
💡 Free Credits: 10,000 characters/month free tier (≈10 minutes), no credit card required
🤖 Supported Models (6)
Sonic (real-time TTS)Sonic 2 (multilingual, voice cloning)Onyx (STT)Mistral (voice design)Pleroma (voice cloning)TTS-Streaming (WebSocket)
✨ Pros
- ✓Sonic model with industry-leading low latency (≈90ms TTFB), the de facto choice for real-time voice agents
- ✓Sonic 2: 29+ languages support, voice cloning, streaming response
- ✓WebSocket streaming interface, purpose-built for real-time voice agents (vLLM-voice alternative)
- ✓Onyx STT high transcription accuracy across 36+ languages
- ✓Pleroma voice cloning: 5-second sample to clone any voice
- ✓MCP protocol support for Claude/Cursor and other AI editor integrations
- ✓State-of-the-art real-time inference latency for latency-sensitive applications
⚠️ Cons
- ×TTS/voice API billing differs from LLM token-based models — cost comparison not straightforward
- ×China access requires stable proxy
- ×Smaller free tier (10,000 characters/month)
- ×Voice quality varies by language (English strongest)
- ×Newer than ElevenLabs, smaller ecosystem and fewer third-party integrations
🎯 Best For
Real-time conversational voice agents; customer service / companion bots; multilingual voice cloning; streaming TTS applications