Researcher

Orhan Firat

Also known as Orhan Fırat

Orhan Firat is a research scientist at Google DeepMind, where he works on large language models. His paper on multi-way, multilingual neural machine translation, written with Kyunghyun Cho and Yoshua Bengio in 2016, was named a best-paper runner-up at NAACL. He then co-authored GShard in 2020, GLaM in 2021, and PaLM in 2022. In December 2023 he was listed as an evaluation co-lead on the Gemini report, and in March 2024 he was among the authors of the Gemini 1.5 report. He oversees the evaluation team of the Gemini program.

Works
127

Publication numbers from OpenAlex, as of 15 October 2026. Citations and h-index are hidden because this OpenAlex record may be incomplete or mix in other authors.

Timeline

  1. 2016

    Multi-way translation at NAACL

    He, Kyunghyun Cho, and Yoshua Bengio publish a paper on multi-way, multilingual neural machine translation in 2016. NAACL names the paper a best-paper runner-up in 2016, and his PhD thesis covers multilingual neural machine translation as well.

  2. 2020

    The GShard scaling system

    GShard, a system for scaling giant models with conditional computation, goes up on arXiv in June 2020 with him as a co-author.

  3. 2021

    The GLaM language model

    In a paper posted to arXiv in December 2021, he co-authors GLaM, a mixture-of-experts language model.

  4. 2022

    The PaLM language model

    PaLM, a 540-billion-parameter language model, goes up on arXiv in April 2022, and he is one of its co-authors.

  5. December 2023

    Evaluation co-lead for Gemini

    The Gemini report, which goes up on arXiv in December 2023, lists him as an evaluation co-lead.

  6. March 2024

    Author of the Gemini 1.5 report

    He appears among the authors of the Gemini 1.5 report, which goes up on arXiv in March 2024.

  7. As of October 2026

    Research scientist at Google DeepMind

    He works as a research scientist at Google DeepMind on large language models, and he oversees the evaluation team of the Gemini program.

Sources and credits

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

  1. Multi-Way, Multilingual Neural Machine Translation with a Shared Attention Mechanism - ACL Anthology aclanthology.org. Checked 15 October 2026.
  2. [1601.01073] Multi-Way, Multilingual Neural Machine Translation with a Shared Attention Mechanism arxiv.org. Checked 15 October 2026.
  3. Best paper runner-up at NAACL’16 – Kyunghyun Cho kyunghyuncho.me. Checked 15 October 2026.
  4. Orhan Fırat orhanfirat.com. Checked 15 October 2026.
  5. [2006.16668] GShard: Scaling Giant Models with Conditional Computation and Automatic Sharding arxiv.org. Checked 15 October 2026.
  6. [2112.06905] GLaM: Efficient Scaling of Language Models with Mixture-of-Experts arxiv.org. Checked 15 October 2026.
  7. [2204.02311] PaLM: Scaling Language Modeling with Pathways arxiv.org. Checked 15 October 2026.
  8. Gemini: A Family of Highly Capable Multimodal Models arxiv.org. Checked 15 October 2026.
  9. [2312.11805] Gemini: A Family of Highly Capable Multimodal Models arxiv.org. Checked 15 October 2026.
  10. [2403.05530] Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context arxiv.org. Checked 15 October 2026.
  11. Turkish InfoTech Society, AMA with Orhan Firat, Senior Staff Research Scientist @ Google DeepMind | LLM, AI, NLP · Luma luma.com. Checked 15 October 2026.