AI Models & Agents
ElevenLabs Debuts Realtime Speech-to-Speech Models
The company says its Mini and Max models handle spoken input and responses in more than 90 languages.
Illustration of a person speaking into a microphone as an abstract voice-agent interface displays audio waveforms.
Photo: Kronos News
ElevenLabs announced Realtime v1 Mini and v1 Max on October 20, describing them as speech-to-speech models designed to power voice agents. The company says both support more than 90 languages, with Mini aimed at lower-cost conversations and Max at more complex tasks. [1]
ElevenLabs says Artificial Analysis independently evaluated the models on public benchmarks. In the company’s account, Max scored 98.8% on Big Bench Audio’s spoken-reasoning questions and began speaking in an average of 0.57 seconds. These are benchmark results reported in the company’s announcement; they do not establish performance in real-world deployments. [1]
Editorial notes
Transparency note
AI assisted drafting. Human edited and reviewed.
- AI assisted
- Yes
- Human review
- Yes
- Last updated
Risk assessment
The second supplied source could not be accessed for verification.
Sources
Related stories
View all
Topics
About the author
Kronos News Desk covers ai models & agents and editorial analysis for Kronos News.