The company says both support more than 90 languages and can use its library of over 17,000 voices. They are available through ElevenAgents and a Realtime API compatible with the OpenAI Realtime specification; the API connects over WebSocket.
ElevenLabs reports that Artificial Analysis evaluated the models on Big Bench Audio, Full-Duplex-Bench, and Speech Agent Arena. On Big Bench Audio, v1 Max scored 98.8% and Mini 75.9%; average time to first audio was 0.57 seconds for Max and 0.62 seconds for Mini. On tool-use tasks, Max completed 87.1% and Mini 60.3%. These are the company’s reported results from the evaluation, not evidence of performance in every deployment.