Eleven v4 launches as #1-ranked emotive voice model

ElevenLabs frames Eleven v4 and Eleven v4 Turbo as the culmination of its latest research in expressive speech generation, built for content where delivery matters as much as the words. The company positions the models against a specific trade-off it says has shaped the market: high-quality voice models have tended to be slower, pushing many builders toward proficient but monotone agents. Eleven v4 Turbo is aimed squarely at that gap, pairing speed with emotion, and it is optimized together with ElevenLabs' ElevenAgents platform as one system rather than stitched from separate vendors.

Beyond delivery, the launch addresses consistency and multilingual reach. Eleven v4 uses a new method for capturing speakers' identities, preserves voice consistency across generations and long-form projects like audiobooks and ads, and lets a voice recorded in one language speak any other fluently while adopting a native accent. Instant Voice Clones now need just 10 seconds of audio, and Professional Voice Clones are supported for high-fidelity needs. Inline tags and natural-language direction prompts, from [laughs] to [said angrily in French accent], are followed more accurately than in prior models.

The stakes hinge on whether the blind-test advantage holds once customers can run the models themselves. The #1 ranking and ~75% listener preference come from September 2026 tests against named competitors including Cartesia Sonic 3.6, Inworld TTS-2, and Google Gemini models, so independent replication and real-world deployments in agents, dubbing, and audiobooks will determine whether ElevenLabs' emotive claim translates into lasting adoption.