ElevenLabs makes Realtime v1 Mini and Max voice AI models available

What is changing?

ElevenLabs has made Eleven Realtime v1 Mini and v1 Max available in an announcement first published on October 20 and updated on October 29. Mini is positioned for natural conversation, educational assistants and interactive characters, while Max targets customer support, lead qualification and multi-step tool use. The company says both models support more than 90 languages and over 17,000 voices from its library.

The models combine speech recognition, turn-taking, a language model and text-to-speech in an end-to-end tuned cascaded architecture. Because each stage produces text, teams can distinguish a recognition error from a wrong decision and apply guardrails and PII redaction inside ElevenAgents.

Eleven Realtime can be selected in ElevenAgents or accessed through a WebSocket endpoint compatible with the OpenAI Realtime API event structure. WebRTC is not supported at launch, and the standalone API does not automatically include ElevenAgents features such as channels, knowledge bases, testing and monitoring.