ElevenLabs v4 Is Here. These Two Really Don’t Know How to Stop.

The Words Were Fine. The Voice Ruined Everything.

Say “I’m fine.”

Now say it like someone who is absolutely not fine.

Same words. Completely different conversation.

That has always been my problem with artificial voices. They could pronounce the sentence, but sometimes completely miss the reason someone would say it.

The pause. The hesitation. That tiny change in tone that tells you whether someone is excited, bored, joking or about to make your evening considerably more complicated.

ElevenLabs keeps working on that gap.

And honestly, I really like these guys. Especially Mati.

Two friends. Four years. An absurd number.

Mati Staniszewski previously worked at Palantir. His co-founder, Piotr Dąbkowski, worked at Google. They were friends long before ElevenLabs became a company people suddenly wanted to put in every presentation about the future of AI. Their official founder background is worth a look.

According to ElevenLabs, its first funding round in 2022 valued the company at $9 million. In September 2026, a $300 million employee tender offer valued it at $22 billion. That was a transaction allowing employees to sell shares—not $22 billion in revenue, and not a fresh $22 billion investment. The company explains the transaction here.

Still. Nine million to twenty-two billion in four years.

Some of us have browser tabs older than that.

But the valuation isn’t really why I follow them. It’s the pace. The creativity. The feeling that they finish announcing one thing and are already halfway through building the next.

Their marketing team deserves credit, too. To my eye, they understand something plenty of technical companies struggle with: people need to hear what a product makes possible.

A voice demo can do that in seconds.

v4 is about delivery, not just pronunciation

The latest release is Eleven v4, alongside v4 Turbo.

ElevenLabs describes v4 as its most expressive text-to-speech model yet, with more control over emotion, pacing and delivery. Turbo is aimed at situations where responding quickly matters, including conversational applications.

The company also points to strong benchmark results. Fine. Those deserve attention, but I’m not going to turn a launch announcement into a personal verdict that it wins every possible use case.

My test is simpler.

Would I keep listening?

Would this voice improve the story, or would I spend the whole time thinking about the software reading it?

A sentence can be technically perfect and emotionally dead.

That matters in narration, advertising and education. It matters even more when the voice is supposed to have a conversation with you.

Amsterdam is more interesting than another office announcement

ElevenLabs also announced an Amsterdam office and an expanded Netherlands team.

Normally, an office announcement would not get me particularly excited. Congratulations on the chairs.

But one detail caught my attention.

The company says its work with KPN initially ran into difficulties recognising Dutch postcodes and dates of birth. Improving performance required work on those local details.

That is the difference between an impressive demonstration and something a customer can actually use.

Nobody calling about their internet connection cares how beautifully the agent delivers its opening sentence if they have to repeat their postcode five times.

A convincing voice gets your attention. Handling the boring details earns your patience.

The startup offer made me pay attention, too

Inside the Claude Startup Stack, the ElevenLabs offer stood out to me: 12 months free and 33 million characters.

That feels genuinely generous.

ElevenLabs’ own Startup Grants announcement describes a 12-month allowance of 33 million credits, also referring to characters of audio-generation credits. Eligibility and the actual grant terms matter; this is not unlimited access or 33 million dollars.

But a year gives a founder something useful: time to experiment without every unsuccessful attempt immediately becoming a small financial argument.

To me, it is also very good marketing.

Give people enough room to build something meaningful, and your product can become part of their business before you need to deliver another sales pitch.

Start with the thing you want to make

For FreeMalta, this isn’t an abstract question. We’ve been considering ElevenLabs for Lore, because some stories deserve to be heard, not just read.

But I don’t want a voice mechanically reading a page. I want the pauses, the tension, the moment you lean in because you need to know what happens next.

Whether we can get that right is something we still need to test. That’s the part that interests me: giving a story another way to reach someone.

It’s where that voice would make the experience better.

An article someone can listen to. A multilingual explanation. A useful conversation instead of another form.

Those are different projects. If you’re weighing narration against an on-screen presentation, our ElevenLabs vs HeyGen DUEL is a useful place to start. If you’re thinking about customer workflows, explore Serve & Automate.

And if you want the background behind the product, there’s our ElevenLabs company story.

I like the founders. I like the creativity. I like how they keep moving.

That doesn’t mean every business needs an AI voice tomorrow.

But if you have something worth saying, take a closer look at ElevenLabs.

Then listen carefully.

The words are only half the job.