AI Models & Platforms
OpenAI to Award 50,000 ChatGPT Plus Grants to Researchers

On November 16, 2026, OpenAI announced that it will award 50,000 new ChatGPT for Academic Researchers grants, with applications opening November 23, 2026. Each grant provides free ChatGPT Plus access for a recipient and up to two collaborators, covering as many as 150,000 researchers this round.
Selection will occur on a first-come, first-served basis among eligible applications, replacing the pilot’s lottery. Qualified researchers already on the waitlist will receive activation instructions by email before applications open. OpenAI’s overall goal is to reach 200,000 researchers, twice the target set when the pilot launched in July 2026.
ChatGPT Plus Replaces Pro for New Grants
The pilot has enrolled 20,000 researchers across 33 countries. Pilot participants received ChatGPT Pro. OpenAI said its usage review found that most participants would have remained within Plus limits, and it attributed the switch to Plus for new grants to its aim of supporting more research teams.
Program participants can use OpenAI’s frontier models and receive training on using ChatGPT in research. Data shared through the program is not used to train OpenAI models by default.
Pilot projects illustrate different research uses. Shuqi Li of Brigham and Women’s Hospital and Harvard Medical School is exploring whether ChatGPT can connect laboratory and patient data to accelerate antibiotic development for tuberculosis. At the Florida Museum of Natural History, Arthur Porto and his team use AI tools to turn specimen photographs into searchable databases. They also created a 3D butterfly model whose wings change color as it is turned, like the real specimen.
How Scientists Used ChatGPT in the Pilot
A companion study from OpenAI’s Economic Research team, Science at the Frontier: How Scientists Use ChatGPT, analyzed 580,186 scientific messages from 1,819 scientists between August 20 and September 22, 2026, alongside a survey of 251 participants conducted from September 30 to October 7, 2026. To protect privacy, automated classifiers sorted messages by research activity; the research team saw the categories but did not read the underlying messages or ChatGPT’s responses. Activity shares were calculated for each scientist and then averaged, giving every scientist equal weight.
The study reports that reading and learning accounted for an average 28% of scientific messages and writing and communicating 27%, followed by analyzing evidence and theories at 17% and writing research software at 14%. Sixty-eight percent of scientists used ChatGPT to develop research ideas, defined as sending at least three ideation messages during the study. Among ideation messages, 36% involved a scientist bringing an idea for ChatGPT to refine, 35% began with a scientist’s specific question, and 23% asked ChatGPT to suggest a question.
The study also measured follow-up requests to clarify, check, or revise ChatGPT’s responses. Average follow-up rates were highest for writing and communicating at 38%, followed by writing research software and analyzing evidence and theories at 34% each. The metric captures a follow-up in the next scientific message in the same conversation within seven days.
Surveyed participants estimated that completing the same work at the same quality without AI would have taken 1.9 times as long. That figure is self-reported. Ninety-eight percent of the 250 scientists answering a separate discovery-timeline question expected AI to accelerate discovery in their field. Those responses describe expectations rather than observed discovery gains.
OpenAI said the findings do not establish causal effects and that determining whether these expectations translate into faster discovery will require more research. The study also identifies funding, staffing, and experimental capacity as constraints on testing ideas. The expansion therefore pairs broader access with an unresolved scientific question: the pilot documents how AI fits into research work, while the larger program’s impact depends on researchers’ ability to test AI-assisted ideas and validate any resulting findings.












