Gradium, a Paris-based company developing the world’s most advanced real-time AI voice technology, has launched after 3 months of stealth. The company builds on more than a decade of frontier research by the founders of Kyutai, a leading non-profit AI research lab. Gradium has raised a $70M seed round led by FirstMark Capital and Eurazeo, with significant participation from DST Global Partners, Eric Schmidt (former CEO and Chairman of Google), Xavier Niel (Iliad), Rodolphe Saadé (CMA CGM), Korelya Capital, Amplify Partners and other investors.
Gradium develops audio language models, an audio-native counterpart to large language models designed to deliver natural, expressive, ultra-low latency voice interactions at scale and capable of performing any voice task. Audio language models were originally invented by Gradium’s founders and have since become the dominant paradigm across the industry; today, the team continues to push the frontier of what these models can do. Gradium’s ambition is to become the technical backbone of global voice technology.
Founded in September 2025 by Neil Zeghidour (Meta/Google DeepMind), Olivier Teboul (Google Brain), Laurent Mazaré (Google DeepMind/Jane Street) and Alexandre Défossez (Meta), Gradium represents one of the world’s highest concentration of generative audio expertise ever assembled, translating more than a decade of open research into production-ready systems for developers and enterprises.The team has demonstrated exceptional execution velocity generating its first revenue within weeks, onboarding early adopters across gaming, AI agents, customer care, language learning and healthcare.
“Today, the potential of voice AI is still largely untapped: we’re at the stage where chatbots were before LLMs” said Neil Zeghidour, Founder and CEO of Gradium. “Existing systems remain brittle, costly, and unable to deliver truly natural interactions. At Gradium, our goal is to make voice the primary interface between humans and machines. To achieve this, we’re eliminating the long-standing trade-off between quality and scalability: combining ultra-realistic expressivity, accurate transcription, and ultra-low-latency interactions at a price point that finally makes high-quality voice ubiquitous.”
The platform launches with multilingual support in English, French, German, Spanish, and Portuguese with additional languages in development, and offers flexible plans from developer access to enterprise-scale deployment.
Gradium’s long-term collaboration with Kyutai provides ongoing access to frontier research in generative audio, allowing the team to rapidly transfer fundamental innovations into commercial-grade systems.
For more information, visit gradium.ai