Mikel Artetxe

Mikel Artetxe is a natural language processing researcher and co-founder of the AI startup Reka, with a background in unsupervised machine translation and multilingual sentence embeddings. He did his PhD at the University of the Basque Country from 2016 to 2020 under Eneko Agirre and Gorka Labaka, and his thesis on unsupervised machine translation won the 2020 EurAI Dissertation Award, followed by a 2021 young researchers prize in computer science. From 2020 to 2023, he was a Research Scientist at FAIR (Meta AI), where he coauthored work on mixture-of-experts language models and OPT. In 2023, he became a co-founder of Reka, which described its multimodal Core, Flash, and Edge models in an April 2024 report. He is a co-founder of Reka and an Honorary Researcher at the University of the Basque Country.

Now at: Reka AI and University of the Basque Country
Born: 1992
Works: 111
Badges: Founder
Bridge
Links: Website

Years link to their sources. Publication numbers from OpenAlex, as of 22 October 2026. Citations and h-index are hidden because this OpenAlex record may be incomplete or mix in other authors.

Timeline

2016–2020

PhD with Eneko Agirre

He does his PhD at the University of the Basque Country from 2016 to 2020, advised by Eneko Agirre and Gorka Labaka. Along the way, he interns at FAIR (Meta AI) in 2018, advised by Holger Schwenk, and at DeepMind in 2019, advised by Sebastian Ruder and Dani Yogatama.
2017

Unsupervised neural machine translation

With Gorka Labaka, Eneko Agirre and Kyunghyun Cho, he posts Unsupervised Neural Machine Translation in 2017, a method that trains a translation system on monolingual corpora alone. The system reaches 15.56 BLEU points on WMT 2014 French-to-English translation, and the paper appears at ICLR 2018.
2018

Multilingual sentence embeddings

With Holger Schwenk, he introduces an architecture in 2018 that learns joint sentence representations for 93 languages with a single BiLSTM encoder. A classifier trained on English annotated data alone transfers to any of the 93 languages, and the paper appears in TACL.
2020–2021

Dissertation and young researcher awards

His thesis Unsupervised Machine Translation, supervised by Gorka Labaka and Eneko Agirre, wins the 2020 EurAI Dissertation Award. In 2021, he receives a prize in the Young Researchers in Computer Science category at the research awards of the Spanish scientific society SCIE and the BBVA Foundation.
2020–2023

Research Scientist at FAIR

He works as a Research Scientist at FAIR (Meta AI) from 2020 to 2023. In 2021, he is first author of Efficient Large Scale Language Modeling with Mixtures of Experts, where a 1.1-trillion-parameter MoE model outperforms a compute-equivalent 6.7B dense model, and in 2022 he is an author of OPT, a suite of decoder-only transformers from 125M to 175B parameters.
2023

Co-founding Reka

In 2023, he becomes a co-founder of Reka, an AI startup. He also takes up an Honorary Researcher post at the University of the Basque Country in 2023.
April 2024

Reka Core, Flash, and Edge

Reka releases its report on Reka Core, Flash, and Edge in April 2024, a series of multimodal models trained from scratch that process text, images, video, and audio. He is one of the 25 listed authors, and the report places Core close to the best frontier models on automatic and blind human evaluations.
As of October 2026

Reka co-founder, IXA researcher

He is a co-founder of Reka, an AI startup building multimodal language models, and an Honorary Researcher at the University of the Basque Country, where he works at 20% dedication in the IXA research group. In September 2025, he gives the invited talk at the presentation of the university's HiTZ Chair of Artificial Intelligence and Language Technology in San Sebastián.