Models & LLMs

Elevenlabs Unveils Realtime Speech Models for AI Conversations

Elevenlabs has introduced two new speech-to-speech models designed for real-time AI conversations, with the Max variant achieving high accuracy and speed.

The Decoder · Oct 20, 2026

What happened

  • Elevenlabs' new speech-to-speech models are designed for real-time AI conversations.
  • The Max variant of the model scores high on accuracy and has a fast response time.
  • Both models are available through specific platforms for different use cases.

Why it matters

The release of these models highlights advancements in real-time AI conversations, with a focus on accuracy and speed. The Max variant's high performance suggests potential applications in customer support and complex tasks, while the Mini variant is optimized for cost-effective natural conversations.

The Elephant take

🐘 鼋Elevenlabs is trying to outdo itself with faster, more accurate models, but the real test will be how these models handle real-world interactions. The Max variant's high score is impressive, but it's not clear if it can scale to complex tasks without sacrificing speed.

Who should care

  • AI developers
  • Customer service providers
  • Researchers

What to do next

  1. Test the models for real-world applications
  2. Compare pricing with other AI solutions
  3. Monitor updates on the Elevenlabs platform

Keep in mind

The claims about performance and cost-effectiveness need to be verified in practical use cases beyond benchmark tests.

Read the original reporting at The Decoder ↗