Elevenlabs has introduced two new speech-to-speech models, Eleven Realtime v1 Max and v1 Mini, designed for low-latency, natural AI conversations.
Key facts
- •Elevenlabs released two new speech-to-speech models called Eleven Realtime.
- •The v1 Max model achieved a 98.8 percent score on the Big Bench Audio benchmark.
- •The v1 Max model has a response time of 0.57 seconds.
- •The v1 Mini model is designed for natural conversation and lower costs.
- •The models are available through the ElevenAgents platform and an OpenAI-compatible Realtime API.
Elevenlabs has released two new speech-to-speech models under the Eleven Realtime brand. The models, designated as v1 Max and v1 Mini, are designed to facilitate real-time AI interactions with high accuracy and low latency.
By the numbers
Model Performance and Capabilities
The v1 Max model is designed for complex tasks such as customer support. It achieved a 98.8 percent score on the Big Bench Audio benchmark and features a 0.57-second response time. According to Elevenlabs, no other model currently on the Artificial Analysis leaderboard matches this combination of speed and accuracy. The smaller v1 Mini model is optimized for natural conversations at a lower cost and includes the ability to distinguish between a user finishing a sentence and making filler sounds like 'mm-hmm.'
Availability and Pricing
Both models are accessible via the ElevenAgents platform and an OpenAI-compatible Realtime API. Pricing for the services is structured on a per-minute basis plus additional token costs, with the v1 Max variant priced higher than the v1 Mini.
Advertisement
This article was independently rewritten by ManyPress editorial AI from reporting originally published by The Decoder.
