OpenAI offers 100,000 researchers free access to GPT-5.6

Why it matters

OpenAI is using free model access to make its closed platform part of scientific infrastructure, seeding research workflows and expert feedback around GPT-5.6.

OpenAI, led by co-founder and CEO Sam Altman (@sama), opened applications Wednesday for free access to GPT-5.6 and other frontier models for 10,000 academic researchers, with plans to expand the program to 100,000 accounts through 2027. ChatGPT for Academic Researchers will initially serve scientists, mathematicians and engineers at selected institutions. (x.com)

The first participants will receive access this summer. OpenAI named the Institute for Advanced Study and France's Ecole normale superieure among the institutions where access is already available. Applicants must work at recognized, degree-granting colleges or universities with high research activity, verify their affiliations and describe their active research and intended scientific use. Applications are open through ChatGPT. (openai.com)

The 100,000 figure includes collaborators. An approved researcher can invite as many as four people from the same institution, but every collaborator must verify an academic affiliation and counts toward the program's total number of accounts. That makes the announced figure a measure of individual access rather than 100,000 separately approved research projects. (openai.com)

Participants will receive access across ChatGPT, ChatGPT Work and Codex, including GPT-5.6 Sol Pro at launch, along with expanded deep research, higher usage limits and larger context windows. OpenAI says the workspaces carry business-grade privacy and security protections, with participant data excluded from model training by default. Training and support from specialists familiar with research workflows are also included. (openai.com)

The program puts a distribution mechanism behind a position Altman and OpenAI chief scientist Jakub Pachocki laid out in a June essay: advanced AI needs to reach people and institutions outside the small group building frontier models. Here, that distribution remains inside OpenAI's products. Researchers studying the models themselves will not receive model weights or training data, access that academics have argued is needed for reproducibility, safety research and independent evaluation, Axios reported. (openai.com)

OpenAI released the GPT-5.6 family for general availability on July 9th. Sol is OpenAI's flagship tier, with Terra positioned for lower-cost everyday work and Luna for faster, less expensive requests. The academic program gives researchers access roughly equivalent to a $200-per-month ChatGPT Pro subscription, according to OpenAI's comparison provided to Axios. That figure describes foregone retail pricing rather than OpenAI's cost of serving the accounts. (openai.com)

OpenAI says the program is part of a commitment to spend more than $250 million through 2027 supporting external scientific research and discovery. That commitment also includes NextGenAI, a $50 million initiative for research institutions, and work with the Department of Energy's Genesis Mission to provide frontier AI to national laboratories and universities. (openai.com)

OpenAI already has a large scientific user funnel to draw from. OpenAI estimates that roughly 1.3 million people use ChatGPT for advanced science and mathematics each week, generating about 8.4 million messages. Those are OpenAI's internal usage figures, and the published index behind them extended only through July 21st. (openai.com)

Free access also gives OpenAI a structured group of domain experts testing GPT-5.6 on research problems. OpenAI says participant feedback will help identify where its models work, where they fall short and how the tools should change. The arrangement can seed scientific workflows around ChatGPT, Codex and OpenAI's connectors at the same time that researchers receive access to computing resources many academic budgets cannot support at commercial prices. (openai.com)