Elevenlabs has introduced two speech-to-speech models under the name Eleven Realtime. The pair is designed for real-time AI conversations, with one model aimed at complex work and a smaller option focused on natural exchanges at lower cost.
Two models for different conversation needs
The v1 Mini is positioned for natural conversation. It can distinguish between someone finishing a turn and making a brief acknowledgment such as “mm-hmm.” That distinction matters in spoken interaction: a system that mistakes a listener's acknowledgment for the end of a turn may respond too soon, while one that misses a completed turn may leave a pause.
The v1 Max targets more complex tasks, including customer support. The two models therefore offer different trade-offs for teams building voice-based AI: Mini emphasizes lower cost and conversational flow, while Max is intended for work that may require more capability.
What the Max performance figures indicate
Elevenlabs says the Max variant scores 98.8 percent on the Big Bench Audio benchmark and has a 0.57-second response time. The company also says no model on the Artificial Analysis leaderboard is currently both faster and more accurate.
Those figures frame the product's pitch around both quality and speed. In a spoken conversation, response time affects how quickly an interaction continues, while benchmark performance offers one way to compare model results. The source does not provide equivalent benchmark figures for v1 Mini, so it is not possible from these details alone to compare the two models on those measures.
Access and pricing
Both models are available through the ElevenAgents platform and an OpenAI-compatible Realtime API. That gives developers two ways to access them, depending on how they plan to build or connect a real-time voice experience.
Pricing combines per-minute charges with token costs. The v1 Max costs more than v1 Mini, consistent with the distinction the company makes between the models. The article points readers to the Elevenlabs pricing page for full details, but does not specify the actual rates.
Choosing between Mini and Max
The choice depends on the intended conversation and its cost requirements. A lower-cost option designed to recognize natural turn-taking may suit simpler exchanges, while Max is presented for more demanding applications such as customer support.
For developers, the release offers a pair of models rather than a single setting for every voice application. The stated performance claims, conversational behavior, access routes and pricing structure can help guide an initial choice, though the source provides no further evaluation details or exact pricing figures.