PeopleofSuperintelligence

Researcher

Matthieu Dinot

Matthieu Dinot is an AI scientist at Mistral AI whose work connects reinforcement learning with language, speech and robotics models. He earned a computer science MSc through École Polytechnique's MVA program, held an INRIA research internship from April to August 2023, and coauthored an IJCAI paper on evolutionary algorithms under noise. At Mistral, he coauthored Magistral, Voxtral, Devstral and Ministral 3, along with speech-recognition and text-to-speech models, a robot-navigation model and a multimodal safety classifier. Since 2024, he has worked on code generation and reasoning with reinforcement learning; in August 2026, he coauthored Le Critique on privileged value functions for LLM reinforcement learning.

Works
11
Citations
25
h-index
3
Badges

Publication numbers from OpenAlex, as of 22 October 2026.

Timeline

  1. 2022–2023

    MVA and an INRIA internship

    At École Polytechnique, he completes a computer science MSc through the MVA program. From April to August 2023, he holds a research internship at INRIA.

  2. 2023

    An IJCAI runtime analysis

    In 2023, he coauthors an IJCAI paper with Benjamin Doerr, Ulysse Hennebelle, and Sebastian Will. Doerr and Will are affiliated with École Polytechnique's Laboratoire d'Informatique (LIX). The paper presents the first mathematical runtime analysis of a simple multi-objective evolutionary algorithm on a classic benchmark under noise.

  3. June 2025

    Magistral's reasoning model

    In June 2025, he coauthors Magistral, Mistral AI's first reasoning model, built on the company's own scalable reinforcement-learning pipeline. Magistral Medium is trained on top of Mistral Medium 3 with RL alone; the team open-sources Magistral Small under Apache 2.0.

  4. July–August 2025

    Voxtral and Devstral

    In July 2025, he coauthors Voxtral, which presents Mini and Small as multimodal audio chat models released under Apache 2.0. He coauthors the Devstral technical report in August 2025; it introduces Devstral-Small, a 24B open-source model for code agents.

  5. January 2026

    The Ministral 3 models

    In January 2026, he coauthors Ministral 3, a family of dense language models with 3B, 8B and 14B parameters for applications with limited compute and memory. Each size has base, instruction-tuned and reasoning variants. Cascade Distillation derives the models through iterative pruning and continued training with distillation.

  6. February–March 2026

    Voxtral Realtime and TTS

    In February 2026, he coauthors Voxtral Realtime, a natively streaming speech-recognition model that matches offline transcription quality at sub-second latency. In March 2026, he coauthors Voxtral TTS, an expressive multilingual text-to-speech model that generates natural speech from as little as three seconds of reference audio.

  7. July 2026

    Robostral Navigate and Shieldstral

    In July 2026, he coauthors Robostral Navigate, an 8B vision-language model that predicts navigation waypoints from monocular RGB images across wheeled, legged and aerial robots. He also coauthors Shieldstral, a 3B-parameter policy-adaptive multimodal safety classifier.

  8. Since 2024

    AI scientist at Mistral

    He serves as an AI scientist at Mistral AI on the code generation and reasoning team and works with reinforcement learning. At GenAI Days Paris in April 2026, he presents Devstral 2 and Mistral Vibe and discusses practical RL training pipelines. In August 2026, he coauthors Le Critique with Siddarth Venkatraman and Laurence Aitchison; it proposes Privileged Value Functions for task-relevant token-level feedback and TETHER, a baseline that adaptively interpolates between group-relative and value baselines.

Career

  1. Since 2026Mistral AIAI scientist
  2. 2023InriaResearch intern

Education

  1. 2023École PolytechniqueMSc in computer science

From the biography and OpenAlex affiliations (the years of papers that list the organization), so some years are approximate.

Most-cited works

  1. Runtime Analyses of Multi-Objective Evolutionary Algorithms in the Presence of Noise2023 · International Joint Conference on Artificial Intelligence · 16 citations
  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. Hot off the Press: Runtime Analyses of Multi-Objective Evolutionary Algorithms in the Presence of Noise2024 · Genetic and Evolutionary Computation Conference Companion (GECCO Companion) · 0 citations
  5. Le Critique: Privileged Value Functions for LLM Reinforcement Learning2026 · 0 citations

From OpenAlex, as of 21 October 2026.

  1. The median h-index is 48 at universities and 14 at frontier AI labsIncludes Matthieu Dinot. 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. 16% of AI career moves in the 2020s went from a university to a companyIncludes Matthieu Dinot. We estimate that 16% of AI career moves in the 2020s went from a university to a company; 29% of moves starting at a university went to a company, and 76% ended the university role.
  3. 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.
  4. 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.

Colleagues

People with dated positions at the same organization in the same years.

Sources and credits

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

  1. Matthieu Dinot | alphaXiv alphaxiv.org. Checked 21 October 2026.
  2. Runtime Analyses of Multi-Objective Evolutionary Algorithms in the Presence of Noise | IJCAI ijcai.org. Checked 21 October 2026.
  3. [2305.10259] Runtime Analyses of Multi-Objective Evolutionary Algorithms in the Presence of Noise arxiv.org. Checked 21 October 2026.
  4. Runtime Analyses of Multi-Objective Evolutionary Algorithms in the Presence of Noise (HTML) arxiv.org. 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. [2509.25193] Devstral: Fine-tuning Language Models for Coding Agent Applications arxiv.org. Checked 21 October 2026.
  8. [2601.08584] Ministral 3 arxiv.org. Checked 21 October 2026.
  9. [2602.11298] Voxtral Realtime arxiv.org. Checked 21 October 2026.
  10. [2603.25551] Voxtral TTS arxiv.org. Checked 21 October 2026.
  11. [2607.20785] Robostral Navigate arxiv.org. Checked 21 October 2026.
  12. [2607.25857] Shieldstral arxiv.org. Checked 21 October 2026.
  13. [2608.16739] Le Critique: Privileged Value Functions for LLM Reinforcement Learning arxiv.org. Checked 21 October 2026.
  14. GenAI Days Paris 2026: Recap of Every Talk | AGO useago.com. Checked 21 October 2026.