PeopleofSuperintelligence

Researcher

Romain Sauvestre

Romain Sauvestre is an AI researcher whose coauthored work spans code, vision, reasoning and speech models. They coauthored Code Llama in 2023, Llama 3 and Pixtral 12B in 2024, Magistral and Voxtral in 2025, and Ministral 3 in 2026. Their 2026 work included streaming speech recognition and multilingual text-to-speech. As of October 2026, they are an AI Scientist on Mistral AI's finetuning team.

Works
9
Citations
17
h-index
3

Publication numbers from OpenAlex, as of 22 October 2026.

Timeline

  1. 2023

    Code Llama

    In 2023, they coauthor Code Llama, a family of large language models for code based on Llama 2.

  2. 2024

    Llama 3

    They coauthor Llama 3 in 2024, a family of foundation models that supports multilingual use, coding, reasoning and tool use.

  3. 2024

    Pixtral 12B

    In 2024, they coauthor Pixtral 12B, a 12-billion-parameter multimodal language model trained to understand natural images and documents.

  4. 2025

    Magistral and pure RL

    In 2025, they coauthor Magistral, Mistral AI's first reasoning model, and present its scalable reinforcement-learning pipeline. They describe a stack for exploring pure reinforcement-learning training of large language models.

  5. 2025

    Voxtral audio models

    In 2025, they coauthor Voxtral Mini and Voxtral Small, two multimodal audio chat models trained to understand spoken audio and text documents.

  6. 2026

    Ministral 3

    In 2026, they coauthor Ministral 3, a family of parameter-efficient dense language models designed for compute- and memory-constrained applications. Its 3B, 8B and 14B sizes each have base, instruction-tuned and reasoning variants.

  7. 2026

    Voxtral Realtime and TTS

    They coauthor Voxtral Realtime in 2026, a streaming speech-recognition model that matches offline transcription quality at sub-second latency. They also coauthor Voxtral TTS, an expressive multilingual text-to-speech model that generates natural speech from as little as three seconds of reference audio.

  8. As of October 2026

    AI Scientist at Mistral

    As of October 2026, they are an AI Scientist at Mistral AI and part of its finetuning team.

Most-cited works

  1. Pixtral 12B2024 · 8 citationsWith Alexandre Sablayrolles, Amélie Héliou, Diego de las Casas, Emma Bou Hanna, Nikhil Raghuraman and Pavankumar Reddy Muddireddy
  2. Ministral 32026 · 4 citationsWith Abhinav Rastogi, Alexander H. Liu, Alexandre Sablayrolles, Amélie Héliou, Andy Ehrenberg and Andy Lo
  3. Magistral2025 · 3 citationsWith Alexander H. Liu, Alexandre Sablayrolles, Amélie Héliou, Andy Ehrenberg, Baptiste Bout and Baudouin De Monicault
  4. Robostral Navigate2026 · 0 citationsWith Alexandre Sablayrolles, Amélie Héliou, Baptiste Bout, Baudouin De Monicault, Chris Bamford and Christian Wallenwein
  5. Shieldstral2026 · 0 citationsWith Alexandre Sablayrolles, Amélie Héliou, Anmol Agarwal, Baptiste Bout, Baudouin De Monicault and Chris Bamford

From OpenAlex, as of 21 October 2026.

  1. The median h-index is 48 at universities and 14 at frontier AI labsIncludes Romain Sauvestre. The median h-index is 48 at universities and 14 at frontier AI labs. We compare publishing careers, recent citations, dual appointments and job moves across 2,420 university researchers and 308 frontier-lab researchers.
  2. Companies supply the most prior-work hires at 7 of the 9 frontier AI labsIncludes Mistral AI. Our estimates of the organizations where 550 people worked directly before joining 9 frontier AI labs across 620 career transitions, comparing the main feeder organizations per lab.
  3. Of 943 people shaping AI who worked at Google, 446 work there nowIncludes Mistral AI. Our estimates of where the 943 people shaping AI who worked at Google and Google DeepMind work now: 446 at Google and 452 across 278 other organizations.
  4. 5,000 frontier users would be 38.2% of the core labs' 13,100 reported staffIncludes Mistral AI. 5,000 frontier users would be 38.2% of the core labs' 13,100 reported staff. We test the estimate against 3,913 named model contributors and dated all-staff records across Google DeepMind, OpenAI and Anthropic, plus six additional frontier-grade labs.

Sources and credits

Photo: We have no published photo of Romain Sauvestre yet. Send us a free-licensed photo.

  1. [2308.12950] Code Llama: Open Foundation Models for Code arxiv.org. Checked 21 October 2026.
  2. [2407.21783] The Llama 3 Herd of Models arxiv.org. Checked 21 October 2026.
  3. [2410.07073] Pixtral 12B arxiv.org. Checked 21 October 2026.
  4. Romain Sauvestre csauthors.net. Checked 21 October 2026.
  5. [2506.10910] Magistral arxiv.org. Checked 21 October 2026.
  6. [2507.13264] Voxtral arxiv.org. Checked 21 October 2026.
  7. [2601.08584] Ministral 3 arxiv.org. Checked 21 October 2026.
  8. [2602.11298] Voxtral Realtime arxiv.org. Checked 21 October 2026.
  9. [2603.25551] Voxtral TTS arxiv.org. Checked 21 October 2026.
  10. Romain Sauvestre - Research Scientist Lead at Mistral AI | The Org theorg.com. Checked 21 October 2026.