Eleven v4 Turbo
ElevenLabs · Audio model
Eleven v4 Turbo is the low-latency variant of Eleven v4, aimed at voice agents and live calls. ElevenLabs reports about 100 ms median inference latency and prices it at half the cost of Eleven v4. At the time of checking…
Current
ReleasedSep 28, 2026
Context window—
Input → outputtext → audio
API price$0.04 per 1K characters (Text to Speech API, listed as "v4 Turbo")
Consumer accessAvailable in ElevenAgents, ElevenCreative and the ElevenAPI; usable on the Free plan (10,000 credits/month) and paid plans
Best for
Real-time voice agents and phone bots that need very low latency with expressive speech.
Overview
Eleven v4 Turbo is the low-latency variant of Eleven v4, aimed at voice agents and live calls. ElevenLabs reports about 100 ms median inference latency and prices it at half the cost of Eleven v4. At the time of checking it held the top spot on Artificial Analysis' text-to-speech arena.
Strengths and limits
- Median inference latency of ~100 ms and ~150 ms median time to first speech (vendor-measured)
- Keeps v4 expressiveness and audio tags at half the v4 price
- Over 90 languages; built into ElevenAgents
- Latency figures are vendor-measured with network latency removed
- Closed model, API and app access only
Published benchmark scores
Artificial Analysis Provider Voice Arena (TTS) Elo1330±18 (rank #1)
Source: artificialanalysis.ai
Scores are as published by the source; settings (reasoning effort, tools, attempts) differ between labs, so compare with care. How to read benchmarks →