OpenAI

Greg Brockman, Vicki Cheung, Ludwig Pettersson, Jonas Schneider, John Schulman, Jie Tang, Wojciech Zaremba

OpenAI Gym is a toolkit for reinforcement learning research. It includes a growing collection of benchmark problems that expose a common interface, and a website where people can share their results and compare the performance of algorithms. This whitepaper discusses the components of OpenAI Gym and the design decisions that went into the software.

Ahmed El-Kishky

Aaditya K. Singh, A. Fry, Adam Perelman, A. Tart, A. Ganesh, Ahmed El-Kishky, A. McLaughlin, Aiden Low, AJ Ostrow, Akhila Ananthram, A. Nathan, A. Luo, Alec Helyar, Aleksander Madry, Aleksandr A Efremov, A. Spyra, Alex Baker-Whitcomb, Alex Beutel, A. Karpenko, Aleksandar Makelov, Alexander Neitz, Alexander Z. Wei, Alexandra Barr, Alexandre Kirchmeyer, A. Ivanov, Alexi Christakis, A. Gillespie, Allison Tam, Ally Bennett, Alvin Wan, Alyssa Huang, Amy McDonald Sandjideh, Amy Yang, Ananya Kumar, Andre Saraiva, Andrea Vallone, A.S. Gheorghe, Andrés G. Garcia, Andrew Braunstein, Andrew Liu, Andrew M. Schmidt, Andrey Mereskin, Andrey Mishchenko, Andy Applebaum, Andrew Rogerson, A. Rajan, Annie Y. Wei, Anoop Kotha, Anubhav Srivastava, A. Agrawal, Arun Vijayvergiya, A. Tyra, Ashvin Nair, Avi Nayak, Ben Eggers, Bessie Ji, Beth Hoover, Bill Chen, Blair Chen, Boaz Barak, Borys Minaiev, Botao Hao, Bowen Baker, Brad Lightcap, Brandon McKinzie, Brandon Wang, Brendan Quinn, Brian Fioca, Brian Hsu, Brian Yang, Brian Yu, B. Zhang, B. Brenner, Callie Riggins Zetino, Cameron Raymond, Camillo Lugaresi, C. Paz, Cary Hudson, Cedric Whitney, Chak Li, Charles Chen, Charlotte Cole, Chelsea Voss, Chen Ding, Chen Shen, Chengdu Huang, C. Colby, Chris Hallacy, Chris Koch, Chris Lu, C. Kaplan, Christina Kim, CJ Minott-Henriques, C. Frey, C. Yu, Coley Czarnecki, Colin Reid, Colin Wei, Cory Decareaux, C. Scheau, Cyril Zhang, Cyrus Forbes, Da Tang, D.P Goldberg, Dan Roberts, Dana Palmie, D. Kappler, D. Levine, Daniel Wright, David Leo, David Lin, David Robinson, Declan Grabb, Derek Chen, Derek Lim, D. Salama, Dibyatanoy Bhattacharjee, Dimitris Tsipras, Dinghua Li, Dingli Yu, DJ Strouse, D. Williams, D. Hunn, Edward Bayes, Edwin Arbus, Ekin Akyürek, Elaine Ya Le, E. Widmann, E. Yani, Elizabeth Proehl, Enis Sert, Enoch Cheung, Eric M. Schwartz, Eric Han, Eric Jiang, Eric Mitchell, Eric Sigler, Eric Wallace, Erik Ritter, Erin Kavanaugh, E. Mays, Evgenii Nikishin, Fangyuan Li, F. Such, Filipe de Avila Belbute Peres, Filippo Raso, F. Bekerman, Foivos Tsimpourlas, Fotis Chantzis, Francis Song, Francis Zhang, Gaby Raila, Garrett McGrath, Gary J. Briggs, Gary Yang, Giambattista Parascandolo, Gildas Chabot, Grace Kim, Grace Zhao, G. Valiant, Guillaume Leclerc, Hadi Salman, Hanson Wang, Hao Sheng, Hao-Yang Jiang, Haoyu Wang, Haozhun Jin, Harshit S. Sikchi, Heather Schmidt, Henry Aspegren, Honglin Chen, Huida Qiu, H. Lightman, I. Covert, Ian Kivlichan, Ian Silber, Ian Sohl, Ibrahim Hammoud, Ignasi Clavera, Ikai Lan, Ilge Akkaya, I. Kostrikov, Irina Kofman, I. Etinger, Ishaan Singal, Jackie Hehir, Jacob Huh, Jacqueline Pan, J. Wilczynski, J. Pachocki, James Lee, J. Quinn, J. Kiros, Janvi Kalra, Jasmyn Samaroo, Jason Wang, Jason Wolfe, Jay Chen, Jason Wang, Jean Harb, Jeff Han, Jeffrey Wang, Jennifer Zhao, Jeremy Chen, J. Yang, Jerry Tworek, Jesse Chand, Jessica Landon, Jessica Y. Liang, Ji Lin, Jiancheng Liu, Jianfeng Wang, Jie Tang, Jihan Yin, Joanne Jang, J. Morris, Joey Flynn, Johannes Ferstad, Johannes Heidecke, J. Fishbein, John Hallman, J. Grant, J. Chien, J. Gordon, Jongsoo Park, Jordan Liss, Jos Kraaijeveld, J. Guay, J. Mo, Joshua Lawson, Josh McGrath, Joshua Vendrow, Joy Jiao, J. Lee, Julie Steele, Juliette Wang, Junhua Mao, Kai Chen, Kai Hayashi, Kai Xiao, Kamyar Salahi, Kan Wu, Karan Sekhri, Karan Sharma, Karan Singhal, Karen Li, Kenny Nguyen, Keren Gu-Lemberg, K. King, Kevin Liu, Kevin R. Stone, Kevin Yu, Kristen Ying, Kristian Georgiev, K. Lim, Kushal Tirumala, Kyle Miller, L. Ahmad, Larry Lv, L. Clare, Laurance Fauconnet, Lauren Itow, Lauren Yang, Laurentia M. Romaniuk, Leah Anise, L. Byron, Leher Pathak, Leon Maksin, Leyan Lo, Leyton Ho, Li Jing, Liang Wu, Lianggeng Xiong, Lien Mamitsuka, Lin Yang, Lindsay McCallum, Lindsey Held, L. Bourgeois, Logan Engstrom, Lorenz Kuhn, Louis Feuvrier, Lu Zhang, Lucas Switzer, Lukasz Kondraciuk, Lukasz Kaiser, Manas R. Joglekar, Mandeep Singh, M. Shah, Manuka Stratta, Marcus Williams, Mark Chen, Mark Sun, Marselus Cayton, Martin Li, Marvin Zhang, Marwan Aljubeh, Matt Nichols, Matthew Haines, Max Schwarzer, M. Gupta, Meghan Shah, Melody Y. Huang, Meng Dong, Mengqing Wang, Mia Glaese, Micah Carroll, Michael Lampe, Michael Malek, Michael Sharman, Michael Zhang, Michele Wang, Michelle Pokrass, M. Florian, Mikhail Pavlov, Miles Wang, Ming Chen, Mingxuan Wang, M. Feng, Mo Bavarian, Molly Lin, Moose Abdool, Mostafa Rohaninejad, Nacho Soto, N. Staudacher, Natan LaFontaine, Nathan Marwell, Nelson F. Liu, N. Preston, Nick Turley, Nicklas Ansman, Nicole Blades, Nikil Pancha, Nikita Mikhaylin, Niko Felix, Nikunj Handa, Nishant Rai, N. Keskar, Noam Brown, Ofir Nachum, O. Boiko, O. Murk, Olivia Watkins, Oona Gleeson, Pamela Mishkin, Patryk Lesiewicz, Paul Baltescu, Pavel Belov, Peter Zhokhov, P. Pronin, Phillip Guo, Phoebe Thacker, Qi Liu, Qim-ing Yuan, Qinghua Liu, Rachel Dias, Rachel Puckett, Rahul K. Arora, R. Mullapudi, Raz Gaon, Reah Miyara, Ren Song, Rishabh Aggarwal, R. Marsan, Robel Yemiru, Robert Xiong, Rohan Kshirsagar, Rohan Nuttall, Roman Tsiupa, Ronen Eldan, Rose Wang, Roshan James, R. Ziv, Rui Shu, Ruslan T. Nigmatullin, Saachi Jain, Saam Talaie, S. Altman, Sam Arnesen, Sam Toizer, S. Toyer, Samuel Miserendino, S. Agarwal, Sarah Yoo, S. Heon, Scott Ethersmith, S. Grove, Sean Taylor, Sébastien Bubeck, Sever Banesiu, Shaokyi Amdo, Shengjia Zhao, Sherwin Wu, Shibani Santurkar, Shiyu Zhao, Shraman Ray Chaudhuri, Shreyas Krishnaswamy, Shuaiqi Xia, Shuyan Cheng, Shyamal Anadkat, S. Fishman, S. Tobin, Siyuan Fu, Somay Jain, Song Mei, Sonya Egoian, Spencer Kim, S. Golden, SQ Mah, Stephanie L. Lin, S. Imm, Steve Sharpe, Steve Yadlowsky, Sulman Choudhry, Su-hyeon Eum, Suvansh Sanjeev, Tabarak Khan, Tal Stramer, Tao Wang, Tao Xin, Tarun Gogineni, Taya Christianson, Ted Sanders, Tejal Patwardhan, Thomas Degry, Thomas Shadwell, Tianfu Fu, Tianshi Gao, T. Garipov, Tina Sriskandarajah, T. Sherbakov, Tomer Kaftan, Tomoya Hiratsuka, Tongzhou Wang, T. Song, Tony Zhao, T. Peterson, V. Kharitonov, V. Chernova, Vineet Kosaraju, Vishal Kuo, Vitchyr H. Pong, Vivek Verma, V. Petrov, Wanning Jiang, Weixin Zhang, Wenda Zhou, Wen-Li Xie, Wenting Zhan, Wes McCabe, Will DePue, W. Ellsworth, Wulfie Bain, Wyatt Thompson, Xiangning Chen, Xi-Meng Qi, Xin Xiang, Xinwei Shi, Yann Dubois, Yaodong Yu, Yara Khakbaz, Yifan Wu, Yilei Qian, Yin Tat Lee, Yinbo Chen, Yizhen Zhang, Yi-Qu Xiong, Yonglong Tian, Y. Cha, Yu Bai, Yu Yang, Yuan Yuan, Yuanzhi Li, Yufeng Zhang, Yuguang Yang, Yujia Jin, Yun Jiang, Yunyun Wang, Yushi Wang, Yutian Liu, Zach Stubenvoll, Zehao Dou, Zheng Wu, Zhigang Wang

This is the system card published alongside the OpenAI GPT-5 launch, August 2025. GPT-5 is a unified system with a smart and fast model that answers most questions, a deeper reasoning model for harder problems, and a real-time router that quickly decides which model to use based on conversation type, complexity, tool needs, and explicit intent (for example, if you say'think hard about this'in the prompt). The router is continuously trained on real signals, including when users switch models, preference rates for responses, and measured correctness, improving over time. Once usage limits are reached, a mini version of each model handles remaining queries. This system card focuses primarily on gpt-5-thinking and gpt-5-main, while evaluations for other models are available in the appendix. The GPT-5 system not only outperforms previous models on benchmarks and answers questions more quickly, but -- more importantly -- is more useful for real-world queries. We've made significant advances in reducing hallucinations, improving instruction following, and minimizing sycophancy, and have leveled up GPT-5's performance in three of ChatGPT's most common uses: writing, coding, and health. All of the GPT-5 models additionally feature safe-completions, our latest approach to safety training to prevent disallowed content. Similarly to ChatGPT agent, we have decided to treat gpt-5-thinking as High capability in the Biological and Chemical domain under our Preparedness Framework, activating the associated safeguards. While we do not have definitive evidence that this model could meaningfully help a novice to create severe biological harm -- our defined threshold for High capability -- we have chosen to take a precautionary approach.

Tianyang Zhong, Zheng Liu, Yi Pan, Yutong Zhang, Yifan Zhou, Shizhe Liang, Zihao Wu, Yanjun Lyu, Peng Shu, Xiaowei Yu, C. Cao, Hanqi Jiang, Hanxu Chen, Yiwei Li, Junhao Chen, Huawen Hu, Yihe Liu, Huaqin Zhao, Shaochen Xu, Haixing Dai, Lin Zhao, Ruidong Zhang, Wei Zhao, Zhenyuan Yang, Jingyuan Chen, Peilong Wang, Wei Ruan, Hui Wang, Huan Zhao, Jing Zhang, Yiming Ren, S. Joe Qin, Tong Chen, Jiaxi Li, Arif Hassan Zidan, Afrar Jahin, Minheng Chen, Sichen Xia, J. Holmes, Zhuang Yan, Jiaqi Wang, Bocheng Xu, Weiran Xia, Jichao Yu, Kaibo Tang, Ya-Xiong Yang, Bolun Sun, Tao Yang, Guoyu Lu, Xianqiao Wang, L. Chai, He Li, Jin Lu, Lichao Sun, Xin Zhang, Bao Ge, Xin-Yue Hu, Lian-Cheng Zhang, Hua Zhou, Lu Zhang, Shu Zhang, Ning-Ning Liu, Bei Jiang, Ling-Yi Kong, Zhen Xiang, Yu-Qing Ren, Jun Liu, Xi Jiang, Yu Bao, Wei Zhang, Xiang Li, Gang Li, Wei Liu, Dinggang Shen, Andrea Sikora, Xiaoming Zhai, Dajiang Zhu, Tianming Liu

This comprehensive study evaluates the performance of OpenAI's o1-preview large language model across a diverse array of complex reasoning tasks, spanning multiple domains, including computer science, mathematics, natural sciences, medicine, linguistics, and social sciences. Through rigorous testing, o1-preview demonstrated remarkable capabilities, often achieving human-level or superior performance in areas ranging from coding challenges to scientific reasoning and from language processing to creative problem-solving. Key findings include: -83.3% success rate in solving complex competitive programming problems, surpassing many human experts. -Superior ability in generating coherent and accurate radiology reports, outperforming other evaluated models. -100% accuracy in high school-level mathematical reasoning tasks, providing detailed step-by-step solutions. -Advanced natural language inference capabilities across general and specialized domains like medicine. -Impressive performance in chip design tasks, outperforming specialized models in areas such as EDA script generation and bug analysis. -Remarkable proficiency in anthropology and geology, demonstrating deep understanding and reasoning in these specialized fields. -Strong capabilities in quantitative investing. O1 has comprehensive financial knowledge and statistical modeling skills. -Effective performance in social media analysis, including sentiment analysis and emotion recognition. The model excelled particularly in tasks requiring intricate reasoning and knowledge integration across various fields. While some limitations were observed, including occasional errors on simpler problems and challenges with certain highly specialized concepts, the overall results indicate significant progress towards artificial general intelligence.

André Brown