ElevenLabs has just dropped a bombshell in the world of AI voice technology with the release of Eleven v3 (alpha) on June 6, 2025—touted as the best model for voice generation to date. This groundbreaking update, available now, brings an unprecedented level of expressiveness and flexibility, leaving users stunned by its realism and capabilities. Whether you’re a content creator, developer, or just curious about AI, this release is turning heads. Ready to explore? Check it out at https://elevenlabs.io/v3.
Table of contents
Open Table of contents
A Leap in Expressiveness with Audio Tags
The star of Eleven v3 is its ability to interpret audio tags, a feature that lets users fine-tune the emotional tone and delivery of generated voices. By embedding tags like [sad], [angry], [happily], [whispers], or [shouts] directly into your text, you can guide the AI to match the mood or intensity you’re aiming for—think whispering a suspenseful line or screaming a dramatic reveal. This level of control builds on ElevenLabs’ reputation for lifelike speech, which has already powered audiobooks, games, and films, and takes it to new heights with context-aware adjustments.
Beyond emotions, v3 introduces the ability to generate multi-speaker dialogues with an unlimited number of voices. This means you can craft realistic conversations where characters interrupt each other or shift tones seamlessly, a feature perfect for storytelling or interactive media. Plus, the model now generates two audio outputs simultaneously, giving you options to choose from or blend, speeding up the process of landing on the perfect result.
Accessibility for All with a Generous Offer
What’s truly exciting is the accessibility. Eleven v3 is available even to free users, who receive 10,000 credits to experiment with—enough for significant testing without upfront cost. For those wanting more, ElevenLabs is offering an 80% discount on v3 generations throughout June 2025, making it an affordable time to dive into this advanced model. This move contrasts with the typical paywalls of high-end AI tools, suggesting ElevenLabs is keen to broaden its user base while the model is in its alpha phase.
The catch? As an alpha release, v3 requires more prompt engineering than its predecessors (like v2.5 Turbo), and it’s not optimized for real-time use yet—best suited for pre-recorded content. The dual-output feature and unlimited speakers also hint at ongoing refinement, but the potential is clear, especially with its support for 70+ languages.
Why This Matters and What’s Next
Eleven v3’s release challenges the narrative that top-tier AI voice tools are reserved for enterprise budgets. By offering credits and discounts, ElevenLabs is democratizing access, which could spark a wave of creative projects from indie developers to educators. The dialogue feature, in particular, could transform how we produce podcasts, games, or dubbed content, though its reliability might vary as it’s still in testing.
Some might question the stability of an alpha model, and posts found on X reflect a mix of awe at its expressiveness and caution about occasional inconsistencies. ElevenLabs acknowledges this, promising continued fine-tuning to improve control and reliability. With a public API on the horizon and the current discount, now’s the time to experiment and see how this tool can elevate your audio projects.
Dive In and Shape the Future of Voice
Eleven v3 (alpha) isn’t just an upgrade—it’s a glimpse into the future of voice generation, blending creativity with accessibility. Whether you’re crafting a dramatic audiobook or a multi-character game dialogue, this model’s expressive power and generous trial offer make it worth a try. Head to https://elevenlabs.io/v3, grab your 10,000 credits, and start exploring. With June’s discount, there’s no better time to see what this unreal AI can do! 😮