[ElevenLabs Bundles Voice Agent: Supports Over 90 Languages, Compatible with OpenAI Interface] Beating AI News Flash: ElevenLabs has released its real-time voice model, Eleven Realtime, available in two versions: Mini and Max. Mini focuses on low-cost conversations, while Max is designed for customer service and complex tasks requiring continuous tool invocation. Both versions support over 90 languages and more than 17,000 voice tones. In Artificial Analysis's 1,000 voice reasoning tasks, Max achieved an accuracy rate of 98.8%, with an average response time of 0.57 seconds to start speaking. On the same leaderboard, Qwen Audio 3.0 Realtime Plus had a slightly higher accuracy rate of 99.2% but required an average of 1.54 seconds to begin speaking. Building a voice agent typically requires independently connecting voice recognition, large models, and voice synthesis, as well as fine-tuning pauses, turn-taking, and interruptions. Eleven Realtime optimizes these processes into a unified system, eliminating the need for piecemeal assembly. The underlying system still connects multiple models in series, and developers cannot individually select models when using the Realtime mode. The new model is being gradually rolled out through ElevenAgents and the Realtime API. The API is compatible with OpenAI's WebSocket protocol, allowing developers to reuse existing SDKs. Projects using WebRTC will need to adjust their connection methods first. [Original Link]