The classic engine on Wixzel Voice: Deepgram + GPT-6 Luna + ElevenLabs
The default pipeline for English. Deepgram Nova-3 transcribes, GPT-6 Luna via OpenRouter reasons, and ElevenLabs Flash v2.5 speaks, in any of its 32 languages. It is the cheapest composed engine and the one to start with.
Last updated
What does each model in the classic engine cost on Wixzel Voice?
A composed engine is three stages, each billed in its own unit. The per-minute column is what that unit adds up to at typical speech rates. The pipeline label is Deepgram + GPT-6 Luna + ElevenLabs.
| Stage | Model | Unit price | Per minute |
|---|---|---|---|
| Speech-to-text | deepgram/nova-3 | $0.0002 per second | $0.0108 |
| Speech-to-text | elevenlabs/scribe_v2_realtime | $0.0002 per second | $0.0091 |
| Language model | openrouter/openai/gpt-6-luna | $0.0001 per 1k tokens in | $0.0003 |
| Language model | openrouter/openai/gpt-6-luna | $0.0007 per 1k tokens out | $0.0003 |
| Text-to-speech | elevenlabs/eleven_turbo_v2_5 | $0.0700 per 1k characters | $0.0630 |
| Text-to-speech | elevenlabs/eleven_flash_v2_5 | $0.0700 per 1k characters | $0.0630 |
| Text-to-speech | elevenlabs/eleven_v4_turbo | $0.0560 per 1k characters | $0.0504 |
| Orchestration | wixzel/orchestration | $0.0120 per minute | $0.0120 |
Rounded up, never down. Where two models are listed for one stage the engine picks by language; the engine price above assumes the dearer path. All prices on the model price list.
Frequently asked questions
What is the classic voice engine?
How much does the classic engine cost per minute?
Which models does the classic engine use?
deepgram/nova-3, elevenlabs/scribe_v2_realtime, openrouter/openai/gpt-6-luna, elevenlabs/eleven_turbo_v2_5, elevenlabs/eleven_flash_v2_5, elevenlabs/eleven_v4_turbo. Each stage is billed in its own unit: seconds of speech recognised, tokens in and out of the model, and characters synthesised.Do I need my own deepgram account to use classic?
wv_live_… API key and a prepaid balance cover the classic engine and every other engine; there are no provider keys to bring.How do I use the classic engine in a call?
POST /v1/agents, for example the stt, llm and tts refs listed above, then place a call with POST /v1/calls. GET /v1/engines lists what is available right now; an engine whose provider is degraded disappears from that list before calls start failing.Is the classic price per minute a flat rate?
GET /v1/usage/events. A quieter or shorter turn costs less; a long monologue costs more.What do developers build with the classic engine?
English bookings need accurate names and dates; Nova-3 transcription with GPT-6 Luna is the cheapest reliable pairing. Switch the stt language for other languages, or the sarvam engine for Indian ones.
Short, scripted outbound calls are the cheapest workload; the classic engine keeps the per-minute cost lowest for English. Use sarvam for Indian-language lists.
The quickstart prompt builds its first agent on the default engine, which is the cheapest reliable pairing for English. An agent can ask for sarvam for Indian languages or gemini-live for the lowest latency; the engines are listed by a tool, so it never has to guess a model name.
Which guides build on the classic engine?
Connect the Wixzel Voice MCP server to Claude Code or claude.ai, and Claude can build a voice agent, place a call over your own SIP trunk and read you the transcript.
Answer your existing Twilio number with an AI receptionist that books appointments, using a Twilio Elastic SIP trunk and the Wixzel Voice TypeScript SDK.
Upload leads, run campaigns with an agent you wrote over your own SIP trunk, and read every call's status and transcript, with the Wixzel Voice API in Node.js.
Which other voice engines does Wixzel Voice run?
Sarvam, Indian languages end to end
Gemini Live, native audio
Deepgram Voice Agent
OpenAI GPT-Live + GPT-6 Luna
OpenAI Realtime, gpt-realtime-2.1-mini
xAI Grok Voice
Name a model for each stage and see the minute it adds up to.
Build your first voice agent