_version_ 1866909943544676352
author SIMA team
Bolton, Adrian
Lerchner, Alexander
Cordell, Alexandra
Moufarek, Alexandre
Bolt, Andrew
Lampinen, Andrew
Mitenkova, Anna
Hallingstad, Arne Olav
Vujatovic, Bojan
Li, Bonnie
Lu, Cong
Wierstra, Daan
Sawyer, Daniel P.
Slater, Daniel
Reichert, David
Vercelli, Davide
Hassabis, Demis
Hudson, Drew A.
Williams, Duncan
Hirst, Ed
Pardo, Fabio
Hill, Felix
Besse, Frederic
Openshaw, Hannah
Chan, Harris
Soyer, Hubert
Wang, Jane X.
Clune, Jeff
Agapiou, John
Reid, John
Marino, Joseph
Kim, Junkyung
Gregor, Karol
Sridhar, Kaustubh
McKinney, Kay
Kampis, Laura
Zhang, Lei M.
Matthey, Loic
Wang, Luyu
Raad, Maria Abi
Loks-Thompson, Maria
Engelcke, Martin
Kecman, Matija
Jackson, Matthew
Gazeau, Maxime
Purkiss, Ollie
Knagg, Oscar
Stys, Peter
Mendolicchio, Piermaria
Hadsell, Raia
Ke, Rosemary
Faulkner, Ryan
Chakera, Sarah
Baveja, Satinder Singh
Legg, Shane
Kashem, Sheleem
Terzi, Tayfun
Keck, Thomas
Harley, Tim
Scholtes, Tim
Roberts, Tyson
Mnih, Volodymyr
Liu, Yulan
Wang, Zhengdong
Ghahramani, Zoubin
author_facet SIMA team
Bolton, Adrian
Lerchner, Alexander
Cordell, Alexandra
Moufarek, Alexandre
Bolt, Andrew
Lampinen, Andrew
Mitenkova, Anna
Hallingstad, Arne Olav
Vujatovic, Bojan
Li, Bonnie
Lu, Cong
Wierstra, Daan
Sawyer, Daniel P.
Slater, Daniel
Reichert, David
Vercelli, Davide
Hassabis, Demis
Hudson, Drew A.
Williams, Duncan
Hirst, Ed
Pardo, Fabio
Hill, Felix
Besse, Frederic
Openshaw, Hannah
Chan, Harris
Soyer, Hubert
Wang, Jane X.
Clune, Jeff
Agapiou, John
Reid, John
Marino, Joseph
Kim, Junkyung
Gregor, Karol
Sridhar, Kaustubh
McKinney, Kay
Kampis, Laura
Zhang, Lei M.
Matthey, Loic
Wang, Luyu
Raad, Maria Abi
Loks-Thompson, Maria
Engelcke, Martin
Kecman, Matija
Jackson, Matthew
Gazeau, Maxime
Purkiss, Ollie
Knagg, Oscar
Stys, Peter
Mendolicchio, Piermaria
Hadsell, Raia
Ke, Rosemary
Faulkner, Ryan
Chakera, Sarah
Baveja, Satinder Singh
Legg, Shane
Kashem, Sheleem
Terzi, Tayfun
Keck, Thomas
Harley, Tim
Scholtes, Tim
Roberts, Tyson
Mnih, Volodymyr
Liu, Yulan
Wang, Zhengdong
Ghahramani, Zoubin
contents We introduce SIMA 2, a generalist embodied agent that understands and acts in a wide variety of 3D virtual worlds. Built upon a Gemini foundation model, SIMA 2 represents a significant step toward active, goal-directed interaction within an embodied environment. Unlike prior work (e.g., SIMA 1) limited to simple language commands, SIMA 2 acts as an interactive partner, capable of reasoning about high-level goals, conversing with the user, and handling complex instructions given through language and images. Across a diverse portfolio of games, SIMA 2 substantially closes the gap with human performance and demonstrates robust generalization to previously unseen environments, all while retaining the base model's core reasoning capabilities. Furthermore, we demonstrate a capacity for open-ended self-improvement: by leveraging Gemini to generate tasks and provide rewards, SIMA 2 can autonomously learn new skills from scratch in a new environment. This work validates a path toward creating versatile and continuously learning agents for both virtual and, eventually, physical worlds.
format Preprint
id arxiv_https___arxiv_org_abs_2512_04797
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle SIMA 2: A Generalist Embodied Agent for Virtual Worlds
SIMA team
Bolton, Adrian
Lerchner, Alexander
Cordell, Alexandra
Moufarek, Alexandre
Bolt, Andrew
Lampinen, Andrew
Mitenkova, Anna
Hallingstad, Arne Olav
Vujatovic, Bojan
Li, Bonnie
Lu, Cong
Wierstra, Daan
Sawyer, Daniel P.
Slater, Daniel
Reichert, David
Vercelli, Davide
Hassabis, Demis
Hudson, Drew A.
Williams, Duncan
Hirst, Ed
Pardo, Fabio
Hill, Felix
Besse, Frederic
Openshaw, Hannah
Chan, Harris
Soyer, Hubert
Wang, Jane X.
Clune, Jeff
Agapiou, John
Reid, John
Marino, Joseph
Kim, Junkyung
Gregor, Karol
Sridhar, Kaustubh
McKinney, Kay
Kampis, Laura
Zhang, Lei M.
Matthey, Loic
Wang, Luyu
Raad, Maria Abi
Loks-Thompson, Maria
Engelcke, Martin
Kecman, Matija
Jackson, Matthew
Gazeau, Maxime
Purkiss, Ollie
Knagg, Oscar
Stys, Peter
Mendolicchio, Piermaria
Hadsell, Raia
Ke, Rosemary
Faulkner, Ryan
Chakera, Sarah
Baveja, Satinder Singh
Legg, Shane
Kashem, Sheleem
Terzi, Tayfun
Keck, Thomas
Harley, Tim
Scholtes, Tim
Roberts, Tyson
Mnih, Volodymyr
Liu, Yulan
Wang, Zhengdong
Ghahramani, Zoubin
Artificial Intelligence
Robotics
We introduce SIMA 2, a generalist embodied agent that understands and acts in a wide variety of 3D virtual worlds. Built upon a Gemini foundation model, SIMA 2 represents a significant step toward active, goal-directed interaction within an embodied environment. Unlike prior work (e.g., SIMA 1) limited to simple language commands, SIMA 2 acts as an interactive partner, capable of reasoning about high-level goals, conversing with the user, and handling complex instructions given through language and images. Across a diverse portfolio of games, SIMA 2 substantially closes the gap with human performance and demonstrates robust generalization to previously unseen environments, all while retaining the base model's core reasoning capabilities. Furthermore, we demonstrate a capacity for open-ended self-improvement: by leveraging Gemini to generate tasks and provide rewards, SIMA 2 can autonomously learn new skills from scratch in a new environment. This work validates a path toward creating versatile and continuously learning agents for both virtual and, eventually, physical worlds.
title SIMA 2: A Generalist Embodied Agent for Virtual Worlds
topic Artificial Intelligence
Robotics
url https://arxiv.org/abs/2512.04797