Kyutai

Kyutai is a French nonprofit AI research lab founded in November 2023 in Paris by Iliad, CMA CGM, and Schmidt Futures, with a €300 million funding commitment, focused on speech AI, multimodal models, and open-research output.
Kyutai

Kyutai

Kyutai is a French nonprofit artificial intelligence research laboratory founded in November 2023 in Paris by Iliad (the French telecommunications group founded by Xavier Niel), CMA CGM (the French shipping and logistics group founded by Rodolphe Saadé), and Schmidt Futures (the foundation founded by Eric Schmidt). The lab was launched with a €300 million ($330 million) initial funding commitment over five years, with emphasis on open-research output across speech AI, multimodal models, and other areas. As of April 2026, Kyutai is one of the principal European nonprofit AI research labs, with a open-research output including the Moshi full-duplex spoken-conversation model (released July 2024 with industry attention).

At a glance

  • Founded: November 2023 in Paris, France, by Iliad, CMA CGM, and Schmidt Futures.
  • Status: Nonprofit research laboratory.
  • Funding: €300 million ($330 million) initial commitment over five years from Iliad, CMA CGM, and Schmidt Futures.
  • CEO / Lead: Patrick Pérez, Director (since founding). Machine-learning researcher; former senior research leader at Valeo AI.
  • Other notable leadership: Hervé Jegou, Senior Researcher; former Director of Research at Meta AI / FAIR. Edouard Grave, Senior Researcher; former Meta AI / FAIR researcher. Laurent Massoulié, Senior Researcher.
  • Open weights: Yes. Kyutai releases models, code, and training data open-source.
  • Flagship outputs: Moshi (full-duplex spoken-conversation model, July 2024); Helium (compact multilingual language model); Mimi (speech codec); published research output.

Origins

Kyutai was founded in November 2023 in Paris by Iliad (the French telecommunications group, with Xavier Niel founder backing), CMA CGM (the French shipping and logistics group, with Rodolphe Saadé founder backing), and Schmidt Futures (the foundation founded by Eric Schmidt). The €300 million initial funding commitment over five years anchored European nonprofit AI research positioning.

The lab's open-research positioning explicitly distinguished it from peer commercial European AI labs including Mistral AI and Aleph Alpha. The emphasis on speech AI, multimodal models, and open-research output anchored research-program direction through 2024 to 2026.

The 2024 release of Moshi (the full-duplex spoken-conversation model) was the lab's most consequential public-facing milestone. Moshi (released July 2024) was a novel full-duplex audio language model with industry attention as a open-research alternative to commercial spoken-conversation systems.

The 2024 to 2026 period has continued the open-research output including Helium (compact multilingual language model) and Mimi (speech codec).

Mission and strategy

Kyutai's stated mission is to advance fundamental AI research and to release open-research output for the broader research community. The strategic premise reflects European nonprofit research positioning, with emphasis on speech AI, multimodal models, and open-research output.

The strategy combines two threads. First, fundamental research across speech AI, multimodal models, and other areas. Second, open-research output including model weights, code, and training data.

The competitive premise reflects Kyutai's distinct positioning as a French nonprofit AI research lab with European-strategic-investor backing, distinguishing the lab from peer commercial European AI labs.

Distribution channels are predominantly open-research publication and open-weights distribution through Hugging Face and other platforms.

Models and products

  • Moshi. Full-duplex spoken-conversation model. Released July 2024. Open-weights through Hugging Face.
  • Helium. Compact multilingual language model (2B parameters). Open-weights.
  • Mimi. Speech codec. Open-source.
  • Published research output. Across speech AI, multimodal models, and other areas.

Distribution channels are predominantly open-research publication and open-weights distribution.

Benchmarks and standing

Kyutai's evaluation framework focuses on open-research output and industry attention rather than horizontal foundation-model leaderboards. Moshi has been consistently characterized in spoken-conversation-AI industry coverage as a novel full-duplex audio language model approach.

Helium and Mimi have continued open-research output through 2024 to 2026.

Leadership

As of April 2026, Kyutai's senior leadership includes:

  • Patrick Pérez, Director (since founding). Machine-learning researcher.
  • Hervé Jegou, Senior Researcher. Former Director of Research at Meta AI / FAIR.
  • Edouard Grave, Senior Researcher. Former Meta AI / FAIR researcher.
  • Laurent Massoulié, Senior Researcher.
  • Senior research staff across the lab's principal research areas.

Funding and backers

  • Initial commitment (November 2023): €300 million ($330 million) over five years from Iliad, CMA CGM, and Schmidt Futures.

Industry position

Kyutai occupies a distinctive position as one of the principal European nonprofit AI research labs, with Iliad, CMA CGM, and Schmidt Futures backing, open-research output, and the Moshi spoken-conversation model.

Competitive landscape

Outlook

  • Continued open-research output through 2026 to 2027.
  • Continued Moshi, Helium, and Mimi iteration.
  • Continued senior research recruitment.
  • The continued European-strategic-investor commitment trajectory through the remaining €300 million commitment period.

Sources

About the author
Nextomoro

AI Research Lab Intelligence

nextomoro tracks progress for AI research labs, models, and what's next.

AI Research Lab Intelligence

Great! You’ve successfully signed up.

Welcome back! You've successfully signed in.

You've successfully subscribed to AI Research Lab Intelligence.

Success! Check your email for magic link to sign-in.

Success! Your billing info has been updated.

Your billing was not updated.