ElevenLabs launched realtime voice models Mini and Max, with Max scoring 98.8% on Big Bench Audio and a 0.57s first audio response. The Nvidia-backed voice unicorn is already at a $22B valuation after

What happened: ElevenLabs released the Eleven Realtime voice model series, with v1 Mini for natural conversation and v1 Max for customer service and multi-step tasks. Both support over 90 languages and more than 17,000 voices from the ElevenLabs library. Key numbers: Max scored 98.8% on Big Bench Audio, third behind StepAudio 3 Realtime (99.7%) and Qwen Audio 3.0 Realtime Plus Why it matters: The release simplifies voice agent building by unifying underlying components, and ElevenLabs faces competition from OpenAI GPT-Realtime, Google Gemini Live and Alibaba's Qwen-Audio-Realtime series. Coverage: 3 outlets in 24h

Nvidia-backed AI voice unicorn releases two models enabling half-second real-time response.

Nvidia-backed AI voice unicorn releases two models enabling half-second real-time response. — impact context not separately stored; see What changed.