ElevenLabs Takes On OpenAI's Realtime Voice API
ElevenLabs' new realtime voice models lead on speed and reasoning and match OpenAI's API, but trail Google, Grok and OpenAI in head-to-head preference.
Read the storyCompany — ElevenLabs develops AI voice and audio technology for speech generation, dubbing, narration and conversational agents. Its products serve creators, developers and businesses deploying voice experiences.
Get AI Secret’s free daily briefing, including ElevenLabs and the wider AI industry.

ElevenLabs develops AI voice and audio technology for speech generation, dubbing, narration and conversational agents. Its products serve creators, developers and businesses deploying voice experiences.
Sources checked . Selected company background and milestones.
Revenue, funding, product launches and major company events, newest first. Expand earlier milestones for more history.
ElevenLabs introduces v1 Mini and v1 Max speech-to-speech models for voice agents, available through ElevenAgents and a WebSocket Realtime API.
Source: ElevenLabs ↗ElevenLabs says ARR passed $500 million in the first four months of 2026. The same update revises its disclosed year-end 2025 figure to $350 million; ARR is an annualized run rate.
Source: ElevenLabs ↗The Series D supports voice agents and audio research. This financing announcement initially described year-end 2025 ARR as over $330 million; the later May disclosure gives $350 million.
Source: ElevenLabs ↗The round, co-led by a16z and ICONIQ Growth, values ElevenLabs at $3.3 billion and brings disclosed funding to $281 million.
Source: ElevenLabs ↗The company announces an $80 million round and unicorn status, alongside expansion of its voice and dubbing tools.
Source: ElevenLabs ↗ElevenLabs brings its AI voice platform to the public, expanding access to synthesized speech.
Source: ElevenLabs ↗The company starts work on AI audio and multilingual voice technology.
Source: ElevenLabs ↗Briefs, articles and relevant sections, newest first. Explore older coverage below.
More coverage
ElevenLabs' new realtime voice models lead on speed and reasoning and match OpenAI's API, but trail Google, Grok and OpenAI in head-to-head preference.
Read the storyElevenAgents now flags AI callers within seconds so businesses can route bots, block cloned voices and serve humans first, starting with enterprise customers.
ElevenLabs brought ElevenReader to Brazil with 70,000 licensed Portuguese titles from Bookwire Brasil, most never recorded, and Porchat's voice; the full catalog needs Ultra.
ElevenAgents Architect, now in alpha, diagnoses agent problems from transcripts and drafts tested fixes, but nothing goes live without human approval.
Publishers on De Marque's Cantook suite can opt in to ElevenLabs narration and sell titles on ElevenReader, targeting the gap between English and other languages.
University students in the US, Canada, EU, UK and Australia get a free year of ElevenReader Ultra but only three months of the creation tools.
Plus: AI Crosses a Real Line, Brutal Speed Game
Plus: NVIDIA‘s New Agent Stack, IBM Ships a Boring Agent
Artificial Intelligence (AI) voice cloning technology is no longer a futuristic concept—it’s here, and it’s making waves. But not all waves are good.
OpenAI Announces GPT-4.5: A Research Preview of Its Largest AI Model Yet