ElevenLabs发布Realtime两款语音模型
2026年10月20日，ElevenLabs发布语音到语音模型系列Eleven Realtime，包括面向低成本自然对话的v1 Mini，以及面向客服、潜客筛选和多步骤工作流的v1 Max。两款模型可通过ElevenAgents平台和兼容OpenAI Realtime规范的API使用，按音频连接时长加Token用量计费，Max价格高于Mini。据智东西援引Artificial Analysis独立测试，Max在Big Bench Audio语音推理评测中以98.8%正确率位列第三，平均首次音频响应0.57秒，落后于StepAudio 3 Realtime（99.7%）和Qwen Audio 3.0 Realtime Plus（99.2%），Mini为75.9%；ElevenLabs则称Artificial Analysis榜单上没有模型同时更快更准，智东西实测Architect搭建助手时认为语音修改设置方便，但中文口音明显、回复偏长且延迟较高。
ElevenLabs将语音识别、推理和语音生成组合成可直接选用的实时交互方案，以降低语音智能体搭建门槛，但独立测试显示Max在Big Bench Audio仅列第三。
Elevenlabs releases two new speech-to-speech models designed for real-time AI conversations
Matthias Bastian View the LinkedIn Profile of Matthias Bastian
Oct 20, 2026
Elevenlabs released two new speech-to-speech models called Eleven Realtime. The Max variant scores 98.8 percent on the Big Bench Audio benchmark with a 0.57-second response time. Elevenlabs says no model on the Artificial Analysis leaderboard is both faster and more accurate right now.
The smaller v1 Mini is built for natural conversations at lower cost and can tell when someone's done talking or just saying "mm-hmm." V1 Max targets more complex tasks like customer support.
#ElevenLabs#Eleven Realtime v1 Max#Eleven Realtime v1 Mini#Artificial Analysis#ElevenAgents#StepAudio 3 Realtime#Qwen Audio 3.0 Realtime Plus#OpenAI Realtime API#Architect