Elevenlabs releases two new speech-to-speech models designed for real-time AI conversations
decoder ·
Elevenlabs has launched two new speech-to-speech models, Eleven Realtime v1 Max and v1 Mini, specifically engineered for real-time AI conversations. The v1 Max variant achieves a 98.8 percent score on the Big Bench Audio benchmark with an impressive response time of just 0.57 seconds. According to Elevenlabs, this combination of speed and accuracy places it at the top of the Artificial Analysis leaderboard, outperforming all other current models. Meanwhile, the smaller v1 Mini model is designed for cost-effective natural interactions, capable of distinguishing between active listening cues like 'mm-hmm' and actual speech completion. While v1 Mini handles general conversation, v1 Max is optimized for more complex scenarios such as customer support tasks. Both models are accessible via the ElevenAgents platform and an OpenAI-compatible Realtime API. Pricing structures vary by model, with v1 Max incurring higher costs than v1 Mini based on usage minutes and tokens.
decoder