ElevenLabs releases two speech-to-speech models for real-time voice conversations

ElevenLabs has released two speech-to-speech models under the Eleven Realtime name. The larger v1 Max variant targets more complex tasks such as customer support, and it scores 98.8 percent on the Big Bench Audio benchmark with a response time of 0.57 seconds. The company says no model on the Artificial Analysis leaderboard is both faster and more accurate at present.

The smaller v1 Mini is built for natural conversations at a lower cost. It can tell when a speaker has finished talking or is only making a backchannel sound such as "mm-hmm."

Both models can be used through the ElevenAgents platform and an OpenAI-compatible Realtime API. Pricing is charged per minute plus token costs, with v1 Max priced higher than v1 Mini, and full details are on the ElevenLabs pricing page.

Developers building voice agents can test a faster, more accurate speech-to-speech option without rewriting code that already targets the OpenAI Realtime API format. Because billing combines per-minute and token charges, teams should estimate costs for their expected call volume before choosing between Max and Mini.