Diego de las Casas
Diego de las Casas is an AI researcher who has co-authored papers on large language models at Google DeepMind and Mistral AI. He joined DeepMind as a research engineer in 2016, and in 2021 he co-authored Scaling Language Models: Methods, Analysis & Insights from Training Gopher. A paper he co-authored, 'An empirical analysis of compute-optimal large language model training', received a NeurIPS 2022 Outstanding Paper award. He then co-authored Mistral 7B in 2023, Mixtral of Experts in 2024, Magistral in 2025, and Ministral 3 in 2026. As of October 2026, he is an AI scientist at Mistral AI, after earlier work at Google DeepMind.
- Now at
- Mistral AI
- Works
- 24
- Citations
- 4,327
- h-index
- 13
Publication numbers from OpenAlex, as of 15 October 2026.
Timeline
-
2016
A research engineer at DeepMind
He joins DeepMind as a research engineer in 2016, working on deep learning and reinforcement learning.
-
2021
Scaling language models
In 2021, he co-authors Scaling Language Models: Methods, Analysis & Insights from Training Gopher.
-
2022
A NeurIPS Outstanding Paper
A paper he co-authors, 'An empirical analysis of compute-optimal large language model training', receives a NeurIPS 2022 Outstanding Paper award.
-
2023
Co-authoring Mistral 7B
Mistral 7B, a 7-billion-parameter language model, appears in 2023 with him among its co-authors.
-
2024
Mixtral 8x7B mixture of experts
Mixtral of Experts, which introduces Mixtral 8x7B as a sparse mixture-of-experts language model, appears in 2024 with him among its co-authors.
-
2025
Mistral's first reasoning model
Magistral, Mistral's first reasoning model, appears in 2025 with him as a co-author.
-
2026
Ministral 3 language models
In 2026, he co-authors Ministral 3, a family of parameter-efficient dense language models.
-
As of October 2026
An AI scientist at Mistral
He is an AI scientist at Mistral AI, having previously worked at Google DeepMind.
Sources and credits
Photo: We have no published photo of Diego de las Casas yet. Send us a free-licensed photo.
- Diego de las Casas – KHIPU
- [2112.11446] Scaling Language Models: Methods, Analysis & Insights from Training Gopher
- NeurIPS 2022 Awards
- [2203.15556] Training Compute-Optimal Large Language Models
- [2310.06825] Mistral 7B
- [2401.04088] Mixtral of Experts
- [2506.10910] Magistral
- [2601.08584] Ministral 3
- Diego de las Casas — The Who's Who of AI | AI Weekly