Elevenlabs Launches New Real-Time Speech-to-Speech Models

Elevenlabs has launched two new speech-to-speech models named Eleven Realtime. The v1 Max variant achieves a score of 98.8 percent on the Big Bench Audio benchmark, with a response time of just 0.57 seconds. According to Elevenlabs, no model on the Artificial Analysis leaderboard is currently both faster and more accurate.

The Eleven Realtime v1 Max model is positioned at the top left corner of the Artificial Analysis Speech to Speech leaderboard, indicating its combination of high accuracy and low latency. In contrast, the smaller v1 Mini model is designed for more natural conversations at a lower cost and can recognize when a speaker has finished talking or is simply acknowledging with sounds like “mm-hmm.” The v1 Max is tailored for more complex tasks, such as customer support.

Both models are accessible through the ElevenAgents platform and an OpenAI-compatible Realtime API. Pricing is based on usage per minute, with the v1 Max priced higher than the v1 Mini. For further details, users can visit the Elevenlabs pricing page.

Source: The Decoder