Sami Jaghouar

Research

Jaghouar's published work centres on training models across distributed compute. OpenDiLoCo (2024) presented an open-source framework for globally distributed low-communication training, and the INTELLECT-1 technical report (2024) described a model trained on such infrastructure.⁶⁵

Jaghouar is the first named individual author of the INTELLECT-2 report (2025), on a reasoning model trained through globally decentralized reinforcement learning, and a co-author of the INTELLECT-3 technical report (2025). Other co-authored work includes TOPLOC (2025), on verifiable inference, and METAGENE-1 (2025), a metagenomic foundation model.⁷⁸⁹¹⁰

Prime Intellect

Jaghouar's GitHub profile, linked from the Hugging Face account that Hugging Face lists as the author of the INTELLECT-1 technical report, describes Jaghouar's role as leading research at Prime Intellect. The same Hugging Face account is a member of the Prime Intellect organization and publishes models there, including a checkpoint released in October 2026.¹²³

Work and contributions

• INTELLECT model series Open weights

  First author of the INTELLECT-1 technical report (2024) and the first named individual author of the INTELLECT-2 report (2025).⁵⁷

• OpenDiLoCo Published research

  First author of OpenDiLoCo (2024), an open-source framework for globally distributed low-communication training, written with Jack Min Ong and Johannes Hagemann.⁶

Timeline

1. Jul 2024Posted OpenDiLoCo as first author.⁶
2. Dec 2024Posted the INTELLECT-1 technical report as first author.⁵
3. Jan 2025Co-authored the TOPLOC and METAGENE-1 papers.⁹¹⁰
4. May 2025Co-authored the INTELLECT-2 report as its first named individual author.⁷
5. Dec 2025Co-authored the INTELLECT-3 technical report.⁸

Where they work

• Prime Intellect 3 people in the wiki

Related people

• Johannes HagemannAlso at Prime Intellect
• Mika SenghaasAlso at Prime Intellect
• Susan ZhangTraining systems · Language models
• Weizhu ChenLanguage models · Training systems
• Jakub PachockiLanguage models · Reinforcement learning
• Edward BeechingLanguage models · Reinforcement learning
• Noam ShazeerLanguage models · Training systems
• Sara HookerLanguage models · Training systems
• Jonathan FrankleTraining systems · Language models
• Bob McGrewReinforcement learning · Language models