ElevenLabs Launched New V4 Speech Models
The upgraded AI models expand language support to over 90 tongues and enable more precise expressive control.
Updated on Sept. 28, 2026 in Artificial Intelligence
Live Poll
Do you trust the rapid advancement of voice AI technology in your daily life?
ElevenLabs has released its v4 and v4 Turbo speech models, which feature expanded expression controls and require only 10 seconds of audio for voice cloning. These models, which now support more than 90 languages, are aimed at developers building voice agents and audio applications.
Why it matters
The company has scaled significantly, reaching an $11 billion valuation and an annualized revenue run rate exceeding $600 million. This release targets the growing demand for multilingual, expressive voice synthesis within enterprise voice-agent workflows.
The v4 models generate audio in real-time as the language model produces text, while allowing users to stack multiple inline expression tags to adjust delivery based on context.
The players
ElevenLabs
A developer of AI-powered speech synthesis and voice cloning software currently valued at $11 billion.
Sequoia
A venture capital firm that led the most recent $500 million funding round for ElevenLabs.
The details
The v4 system operates by integrating expression controls directly into the inference pipeline, allowing the model to adapt its delivery in sync with text generation. Users can influence prosody and tone by stacking tags that modulate the audio output as the model constructs the response. The platform also enables rapid voice cloning by processing just 10 seconds of source audio to capture a target persona.
Timeline
September 28, 2026: ElevenLabs launched the v4 and v4 Turbo speech models.
The Tech Race
ElevenLabs is positioning its v4 models to capture a larger share of the voice agent market, where corporate adoption currently drives over 55% of its business. This release marks a departure from standard synthetic speech by prioritizing granular expression control alongside expanded global language support.
Developers can now utilize the new v4 models for projects requiring voice cloning with only 10 seconds of audio or multilingual support in over 90 languages. These capabilities are designed for integration into active voice agent workflows, though the specific availability of API tiers was not detailed.
The takeaway
ElevenLabs is moving to solidify its lead in the voice synthesis space by scaling its model complexity to meet enterprise requirements. Observers should track the company's expansion into India, Europe, and Brazil to see if these new language capabilities translate into further market dominance.
Further reading
For more on the latest trends in synthetic speech, visit Artificial Intelligence.
Live Poll
Do you trust the rapid advancement of voice AI technology in your daily life?
