LondonElevenLabs’ new flagship voice model ranks third on voice reasoning.
Artificial Analysis scored it 98.8%, behind StepFun’s 99.7% and Alibaba’s 99.2%.
Eleven Realtime Max answers in 0.57 seconds on average, against 1.54 seconds for Alibaba’s Qwen Audio 3.0 Realtime Plus.
Hands-on testers heard a clear foreign accent in Chinese, and replies ran long.
ElevenLabs’ chief executive wants the company ready to list within 2.5 years.
How each outlet framed it
- Zhidx
- tests the Architect setup assistant hands-on, finding voice edits easy but Chinese accent strong and latency high
Sources: Zhidx