1300 166 138 Tell us what you need
News / OpenAI

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

By Monika Tantau • Published 4 months ago
OpenAI gives 100,000 researchers free access to GPT-5.6

Key takeaways

• OpenAI will give 100,000 researchers free frontier model access, starting with 10,000 seats now
• Each approved workspace gets five seats for 12 months, with Pro level rate limits
• Eligibility needs an arXiv, bioRxiv or ChemRxiv paper from the last three years
• Not all institutions qualify, and Australian researchers must check the list before applying

Most Read

OpenAI has opened applications for ChatGPT for Academic Researchers, a program that hands 100,000 scientists, mathematicians and engineers free access to its frontier models, according to OpenAI. It starts with 10,000 researchers and runs through 2027, and the first workspaces are already live at the Institute for Advanced Study and École normale supérieure.

This is not a discount on ChatGPT. It is a separate workspace with business grade data protections, five seats and Pro level usage limits, free for 12 months. The part that matters for Australian universities is the gate: eligibility runs off a list of selected institutions in supported countries, and not every institution is on it.

What the program actually includes

Approved researchers get the GPT-5.6 family across ChatGPT, ChatGPT Work and Codex, with GPT-5.6 Sol Pro available at launch. OpenAI is also bundling more than 75 life science skills covering genomics, sequencing, single cell analysis and protein modelling, plus connectors into scientific literature and public genomic databases. The differences between the three GPT-5.6 tiers decide which one suits which job.

The capability claims come with numbers. On FrontierMath Tier 4, which tests research level mathematical reasoning, OpenAI reports GPT-5.6 Sol at 83% against 72.5% for GPT-5.5. On GeneBench Pro, which scores complex biological data analysis, Sol Pro solves 31.5% of tasks. Both are OpenAI’s own figures, so treat them as vendor reported until someone repeats them.

The eligibility test is narrow

This is not open to any academic. Applicants must be research faculty or postdoctoral researchers at an eligible institution in a supported country, and they must submit a paper from arXiv, bioRxiv or ChemRxiv posted in the last three years that lists them as an author. The paper has to sit in one of seven fields: biological sciences, chemistry and materials, computer science, earth and planetary sciences, engineering, mathematics or physics.

Verification runs through SheerID, a third party service, using the institutional email that matches the ChatGPT account. OpenAI says review takes one to two weeks. A payment card is required at checkout for verification only, and the total is zero. Meeting the criteria lets you apply; it does not guarantee approval.

The eligibility test is narrow

What happens after 12 months

The workspace does not renew automatically and nobody is charged during the free period. OpenAI says it will share continuation options before the offer ends, and if none is taken the workspace is deactivated after a grace period. Two other limits matter for anyone budgeting around this: the offer covers ChatGPT access and does not include API credits, and the self service plan has no single sign on, no SCIM and no domain claiming.

Why Australian researchers should check the list

Australia is not named in the announcement, and the two launch institutions are American and French. The stated rule is that institutions must be recognised, degree granting universities with a high level of research activity, which describes most of the Group of Eight, but OpenAI still picks the list. Anyone here should check it before spending time on an application.

The plumbing already exists at some Australian universities. UNSW Sydney signed the largest ChatGPT Edu agreement in the country, 10,000 licences covering all fixed term and permanent staff, after a pilot in which 98% of about 800 participants wanted to keep access. Researchers at institutions with Edu still have to apply, but approval is then handled through the university’s own workspace.

A pipeline play, not charity

The program sits inside a commitment OpenAI puts at more than $250 million through 2027, alongside its $50 million NextGenAI initiative and work with the US Department of Energy’s Genesis Mission. OpenAI also published its own usage data: roughly 1.3 million people a week already use ChatGPT for advanced science and mathematics, generating about 8.4 million messages.

There is a commercial logic here that Australian research offices should read plainly. A free year builds habits in the people who later specify tooling for grants, labs and spinouts, and the researchers using AI hardest are already handing it bigger jobs. OpenAI reports that the top fifth of AI users in a field are almost twice as likely to give it tasks estimated at four hours or more.

For a university weighing this against a paid Edu rollout, the question is not whether the free year is worth taking. It is what the second year costs once 12 months of research workflows have been built on it, which is the kind of question our AI consulting work usually starts with.

Scroll to Top