OpenAI Opens 50,000 ChatGPT Grants for Researchers on November 23, First Come, First Served

November 17, 2026 · 5 min read

Granted Research Team · Editorial policy

OpenAI's summer pilot gave academic researchers its most expensive tier and picked recipients by lottery. The second round, opening November 23, 2026, reverses both. Recipients get the cheaper plan, and the first eligible applicants through the door get the grants.

OpenAI announced on November 16 that its ChatGPT for Academic Researchers program will add 50,000 new grants, each covering a researcher and up to two collaborators. That is capacity for up to 150,000 people in this round. Combined with the 20,000 researchers in 33 countries who joined after the pilot launched in July, it moves the program toward a stated goal of 200,000 researchers, according to The Next Web and Semafor. Coverage describing "200,000 free accounts" overstates it: 200,000 is the program target, and this round tops out at 150,000 seats.

For a lab group, the decision is simple: the grant costs nothing. The more interesting questions are what it covers, what it leaves out, and how to be among the first 50,000 eligible applications.

What the Grant Actually Is

The award is a software subscription, not money. Grantees in this round get ChatGPT Plus, not the ChatGPT Pro tier that pilot participants received. TNW called the downgrade "the catch." OpenAI's explanation, as reported, is that pilot usage data showed most participants would have stayed within Plus limits anyway.

That claim is plausible for the typical use pattern, which OpenAI describes in its accompanying study. For heavy users it does not hold. Researchers who leaned on Pro's higher limits for long-context literature synthesis, extended reasoning runs or bulk document analysis will run into caps on Plus that they never saw in the pilot.

Several limits matter for research planning:

First Come, First Served Changes the Strategy

The pilot used a lottery. This round grants access to the first 50,000 applications that meet the eligibility requirements. Neither OpenAI's announcement nor the news coverage gives an opening hour for November 23. Assume the window opens early U.S. time and that demand is front-loaded, because 20,000 researchers signed up for a program that was initially framed as a smaller pilot.

The guide, as summarized, describes intended applicants as research faculty and postdoctoral researchers in eligible regions. Graduate students do not appear to be primary applicants, though they can likely be covered as collaborators on a faculty or postdoc grant. Availability is described as all regions except China and Hong Kong, consistent with OpenAI's general regional availability policy.

Practical preparation for the 23rd:

  1. Pick the applicant now. The person with the clearest eligible appointment (a faculty member or postdoc with an institutional email and a public profile page) should apply. Collaborator slots go to whoever on the team will use the tool most.
  2. Prepare the institutional details. Have your appointment title, institutional email, department URL and a short research description ready to paste. In a first-come queue, anything that sends you looking for information costs you position.
  3. Check the waitlist. One summary reports that people already on the waitlist should receive activation instructions before public applications open. If you signed up during the pilot, look in your inbox, including spam, before applying again.

Read the Data Terms Before You Upload Anything

Coverage of the data policy varies, and the variation matters for anyone working with sensitive material. Semafor reported that OpenAI "will not collect data" from program users. The more precise wording, as cited elsewhere, is that program data is not used for model training by default. That is a narrower commitment. It concerns training, not whether data is processed or retained.

For most literature review, drafting and coding, the default is adequate. For anything covered by an IRB protocol, a data use agreement, HIPAA, FERPA, export controls or a funder's data management plan, it isn't enough to rely on. A free ChatGPT Plus seat is a consumer-tier product, not an institutional enterprise agreement with negotiated terms. Researchers who put identifiable human-subjects data into it may break their own protocols even if OpenAI never trains on the data. Check whether your university already has an enterprise or education agreement with stronger contractual protections, and use that for regulated work.

Federal funders add another layer. NIH has said peer reviewers may not use generative AI tools to analyze or write critiques of applications, because doing so breaches confidentiality. Many agencies also now ask applicants to disclose substantive AI use in proposals. A free seat does not change any of those obligations.

What OpenAI's Own Study Says, and Doesn't

OpenAI paired the announcement with a study of pilot usage, which it calls "Science at the Frontier." It covers more than 580,000 messages from 1,819 scientists, sorted by automated classifiers. On average, reading and learning made up 28 percent of science-related messages and writing made up 27 percent. Surveyed participants estimated the same work would have taken about 1.9 times as long without AI.

Those figures come from a self-selected, self-reported sample of about 250 survey respondents, and OpenAI acknowledges the study does not show that AI speeds up discovery. It names funding, staff and lab capacity as the real constraints. That caveat matters for grant-seekers. A free chatbot subscription may cut the hours spent reading and drafting, but it does not replace the money that buys reagents, compute time, field seasons or a postdoc's salary.

The usage pattern is still a useful guide to what the tool is good for. Literature triage, first drafts of methods and background sections, code scaffolding and turning reviewer comments into revision plans are the tasks where researchers reported the most value. They are also the tasks that take up a large share of the time spent preparing a proposal.

The Wider Race for Academic Users

OpenAI is not alone in this. TNW notes that the company also offers free GPT-Rosalind seats to U.S. university labs and has funded a Simons Foundation effort, while Anthropic and Google DeepMind have launched their own science-access programs this year. The incentive is clear. Researchers who build their habits on one platform during graduate school and their first faculty years tend to bring that platform into the institutional procurement decisions that follow.

For individual researchers, the competition is mostly good news, since there are several overlapping free or discounted offers. Before treating any single one as permanent infrastructure, compare data terms, usage limits and API availability across them, and plan for any free seat to expire.

The new round adds tools, not money: the grants, fellowships and cooperative agreements that fund the actual science still have to be found and won. Researchers looking for those funding opportunities can use Granted to search open calls and draft the proposals that pay for the lab.

Get AI Grants Delivered Weekly

New funding opportunities, deadline alerts, and grant writing tips every Tuesday.

More Tips Articles

Not sure which grants to apply for?

Use our free grant finder to search active federal funding opportunities by agency, eligibility, and deadline.

Find Grants

Ready to write your next grant?

Draft your proposal with Granted AI. Professional members win a grant in 12 months or get a full refund.

Backed by the Granted Guarantee