ElevenLabs Launches v4 Voice Model With Expression Control Across 90+ Languages
ElevenLabs has released Eleven v4, its new flagship text-to-speech model, extending language support to more than 90 languages and adding stronger control over emotional expression, up from the 70 languages supported by its predecessor, Eleven v3. The company describes v4 as its “fastest and most emotive voice model,” pairing high-fidelity voice cloning that preserves a speaker’s characteristics across long text generations with a dedicated Text to Dialogue API for multi-speaker conversations. The release comes as ElevenLabs reports roughly $600 million in annualized recurring revenue as of June 2026, up from $330 million at the end of 2025, following an $11 billion valuation reached in its February 2026 Series D round.
What Happened
Table of Contents
Eleven v4 now sits as ElevenLabs’ top-tier model, replacing v3 as the company’s most capable option for generating synthetic speech. The model’s language list spans major world languages alongside less commonly supported ones — ElevenLabs’ own documentation lists languages from Afrikaans and Amharic through Arabic, Armenian, and Assamese as part of the more than 90 languages covered, reflecting a substantial expansion in linguistic reach aimed at global enterprise deployments rather than English-first use cases.
Alongside the standard v4 model, ElevenLabs also offers a v4 Turbo variant built for real-time interactive applications, achieving roughly 100 milliseconds of latency — fast enough to support live voice agents and conversational applications where delay is noticeable to users.
What’s New in v4
- Language coverage: More than 90 languages, up from 70 in Eleven v3.
- Character limit: Doubled from 5,000 to 10,000 characters per generation.
- Expression control: Higher emotional range and contextual understanding for generating dialogue, with the company citing “highest quality audio, expressiveness, and control” as the core improvement over v3.
- Voice cloning: High-fidelity cloning with reliable speaker preservation across long text generations, reducing the drift in voice character that can occur over lengthy outputs.
- Text to Dialogue API: A dedicated API for generating natural multi-speaker conversations rather than single-voice narration.
- v4 Turbo: A low-latency variant at roughly 100ms response time for real-time voice agent use cases.
Why It Matters
Voice AI has moved from a novelty feature into core enterprise infrastructure. ElevenLabs says 41% of Fortune 500 companies now use its platform, with named clients including The Washington Post, TIME, and HarperCollins, alongside more recent enterprise voice-agent deployments at Deutsche Telekom, Square, and Klarna. Klarna’s implementation is a concrete example of what expressive, low-latency voice models enable at scale: the company put an ElevenLabs voice AI agent in place as first-line phone support for 35 million U.S. customers, reporting up to 10 times faster resolution times as a result.
The expanded language support in v4 is directly relevant to that kind of enterprise deployment, since a voice agent handling customer support at global scale needs to serve callers in far more than English and a handful of major European languages. Going from 70 to more than 90 supported languages closes gaps for companies operating across Africa, South Asia, and other regions with linguistic diversity that earlier voice AI models underserved.
Pricing and Availability
ElevenLabs offers v4 across its existing subscription tiers, from a free plan with 10,000 monthly credits up through a $990-per-month Business plan with 6 million credits and custom Enterprise pricing above that. Paid tiers between the $6 Starter plan and the $299 Scale plan give access to progressively higher credit allowances, with per-minute costs falling from roughly $0.36 on the free tier to around $0.17–$0.18 per minute on Creator-level plans and above. As a launch incentive, ElevenLabs is giving Creator-plan subscribers and above up to double their monthly text-to-speech credits when used specifically on v4, available through the company’s web and mobile apps.
The Competitive Landscape
ElevenLabs competes against both large platform companies and specialized rivals. OpenAI, Google, Meta, and Microsoft have all built voice generation capabilities, typically as a secondary feature inside broader AI product suites rather than a standalone focus. More specialized competitors, including Cartesia and Deepgram, compete directly on speech technology, while ElevenLabs has increasingly differentiated itself through dedicated audio research and proprietary models aimed specifically at voice quality and expressiveness rather than general-purpose AI capability.
The company’s growth trajectory — from a $3.3 billion valuation in January 2025 to $11 billion roughly thirteen months later, with ARR climbing from $330 million to $600 million over the same stretch — suggests that positioning as a focused, best-in-class voice specialist rather than a generalist AI lab has resonated with enterprise buyers even as bigger platforms add their own voice features.
Frequently Asked Questions
How many languages does ElevenLabs v4 support?
Eleven v4 supports more than 90 languages, according to ElevenLabs’ documentation, up from 70 languages supported by the previous Eleven v3 model.
How is v4 different from Eleven v3?
V4 expands language support from 70 to more than 90 languages, doubles the character limit per generation from 5,000 to 10,000, improves emotional expressiveness and voice-cloning consistency over long text, and adds a dedicated Text to Dialogue API along with a low-latency v4 Turbo variant for real-time use.
Who uses ElevenLabs’ voice AI?
ElevenLabs says 41% of Fortune 500 companies use its platform, with named enterprise clients including The Washington Post, TIME, HarperCollins, Deutsche Telekom, Square, and Klarna, which uses an ElevenLabs voice agent as first-line phone support for 35 million U.S. customers.
Related Articles
- Google’s Gemini 4 Enters Post-Training as DeepMind Races to Ship Before Year-End
- Meta Unveils Ray-Ban Meta Gen 3 Smart Glasses Starting at $449
