ElevenLabs

Staniszewski and Dąbkowski are high school friends who grew up together in Poland; Staniszewski worked as a deployment strategist at Palantir and Dąbkowski as a machine learning engineer at Google [8][11]. They were inspired by the "rough" dubbing of American films they watched growing up, specifically Poland's single-narrator voiceover tradition, in which one actor dubs foreign films in monotone for both genders [2][8][18][19]. Their initial goal was to "give machines a voice as natural as our own" [2]. The pair pooled their savings, quit their jobs by May 2022, and personally fronted the first $100,000 training run [18]. The company began fully remote, hiring across Europe and Asia rather than only on the US West Coast [19].

ElevenLabs publicly released its beta platform in January 2023, with a first model that could take text and read it aloud in any voice, including clones [11][18]. It passed one million registered users within months [1]. In August 2023 it expanded voice generation to 28 languages and exited beta [11].

The product line is organized around three platforms. ElevenAgents is an enterprise platform for deploying voice and chat agents at scale, with a visual builder for non-technical users and programmatic control for developers [3][7][8]. ElevenCreative is a no-code browser application for generating and editing speech, music, image, and video across more than 70 languages [5][7][8]. ElevenAPI is a REST interface to the company's foundational audio models, with official Python and TypeScript SDKs [7][8].

Eleven v3, released in June 2025, is described by the company as its most emotionally rich and expressive speech model; it supports more than 70 languages and audio tags such as [excited], [whispers], and [sighs] [6][7][11]. Eleven Multilingual v2 covers 29 languages and is optimized for long-form generation, while Eleven Flash v2.5 offers roughly 75-millisecond latency across 32 languages [6][7]. Scribe v2 provides speech recognition in more than 90 languages, with speaker diarization for up to 32 speakers [7]. The company's models analyze context to detect emotions and adjust intonation and pacing; the technology is in the process of being patented [11].

Instant Voice Cloning requires one to five minutes of audio and targets content creators, while Professional Voice Cloning requires at least 30 minutes of clean audio and targets audiobooks and video games [10]. Users can also generate custom voices from text descriptions [6][10]. The company maintains a voice library of more than 10,000 voices, including those of Michael Caine and Matthew McConaughey [7][18]. Through the Voice Marketplace, creators clone, license, and monetize their voices; the program had paid out $10 million by late 2025 and $22 million by May 2026 [2][3][19].

Eleven Music, launched in August 2025, generates studio-grade music from natural-language prompts and is cleared for commercial use [11]. The model is fully licensed; ElevenLabs spent 18 months negotiating agreements with all four major labels and others such as Kobalt [2][19]. Other releases include AI Dubbing (October 2023), which translates speech into more than 20 languages while preserving the original voice; the Reader app (June 2024); Voice Isolator (July 2024); a conversational AI developer platform (November 2024); and an audiobook publishing platform (February 2025) [11].

ElevenLabs raised a $2 million pre-seed round in January 2023 led by Credo Ventures [11]. A $19 million Series A, announced in June 2023 at a valuation of about $100 million, was co-led by Andreessen Horowitz, Nat Friedman, and Daniel Gross [1][11]. An $80 million Series B in January 2024 valued the company at $1.1 billion [11], and a $180 million Series C in January 2025, co-led by Andreessen Horowitz and ICONIQ Growth, valued it at $3.3 billion [11][15]. An employee tender offer in September 2025 was conducted at a $6.6 billion valuation [5][11]. On February 4, 2026, the company raised a $500 million Series D led by Sequoia Capital at an $11 billion valuation, bringing total funding to $781 million [5]. In May 2026 it added investors including BlackRock, NVIDIA, Jamie Foxx, and Eva Longoria [4], and in June 2026 Poland invested in the company [3]. In March 2026 ElevenLabs said it aims to be IPO-ready within two to three years [16].

ElevenLabs sells subscriptions ranging from a free tier to custom enterprise plans, metered through credits, with text-to-speech billed at one credit per character [7][9]. Revenue is split roughly evenly between corporate customers and creators such as YouTubers, podcasters, and authors [18]. Corporate customers include Cisco, Twilio, and the recruitment agency Adecco [18]. ElevenAgents users include Deutsche Telekom, Square, the Ukrainian government, and Revolut; ElevenCreative users include Duolingo, NVIDIA, and TIME; ElevenAPI users include Meta, Epic Games, and Salesforce [5]. Epic Games used the technology to voice Darth Vader in Fortnite, with the consent of James Earl Jones's estate [18]. Forbes reported that the company charges up to three times as much as American rivals [18].

The company closed 2025 with more than $330 million in annual recurring revenue and crossed $500 million in May 2026 [4][5]. Forbes reported $193 million in trailing twelve-month revenue as of late 2025 and profitability "since earliest days" [18]. In a benchmark by data-training startup Labelbox, ElevenLabs' model made half as many errors as its closest competitor, OpenAI [18]. ElevenAPI powers platforms reaching more than one billion users [5].

In January 2024, robocalls imitating President Joe Biden urged New Hampshire voters to skip the Democratic primary; audio experts concluded the calls were made using ElevenLabs [11][15][18]. Recorded Future reported in December 2024 that the tool had been used in a Russian propaganda operation [15], and Motherboard reported in 2023 that 4chan users had created voices of Joe Rogan, Ben Shapiro, and Emma Watson [15]. In August 2024, voice actors Karissa Vacker and Mark Boyett filed a federal lawsuit in Delaware alleging that ElevenLabs misappropriated their voices to build its text-to-speech software [13][17]. ElevenLabs denied wrongdoing, and the case settled out of court in November 2025 [18].

The company released an AI Speech Classifier in June 2023 to detect whether audio originated from its systems [11], and it follows the C2PA standard for content metadata [15]. Newly cloned voices must pass a consent check, the company maintains a "no go" list of mostly politicians and celebrities, and seven full-time human moderators review clips for misuse [18]. Its Impact program provides free licenses to individuals with accessibility needs and to nonprofits [3][8], and in March 2026 it pledged $1 billion in free voice-restoration technology to one million people living with permanent voice loss [11].

In 2026 the company announced partnerships with the UK government to bring voice AI to public services alongside a London headquarters expansion [3][4], with Hasbro to add its intellectual property to the Voice Library [3], and with LOT Polish Airlines [3]. It released an audiobook of The Odyssey narrated by Michael Caine's AI voice [3], and expanded in Spain, California (creating 173 jobs), and Canada [3][4]. The ElevenLabs Summit 2026 was held in Warsaw in June 2026, and Alex Holt was appointed Field CTO that month [3][4].