OpenAI launched Prism on Tuesday, a free LaTeX-based web workspace that embeds ChatGPT (GPT-5.2) into scientific writing to speed drafting, revision, and collaboration, aiming to fold its models into high-value research (see the announcement). The company says the tool integrates advanced reasoning into in-place drafting and citation workflows to accelerate scientific work.
Prism is built on the LaTeX platform Crixet, which OpenAI acquired earlier this month (Crixet), and the app supports typesetting, citations, and collaborative editing with model assistance. OpenAI CEO Sam Altman discussed the launch in a town hall, noting internal feedback that the models are producing nontrivial scientific progress, and adding “With 5.2, a special version we use internally, we’re now for the first time hearing from scientists that the scientific progress of these models is no longer super trivial.”
Jonathan Schaeffer, a professor emeritus and co-founder of Synsira, said Prism helps writing and literature searches but warned about using models for core research inference. He observed “There are two issues with writing papers,” and predicted “hallucinations will not go away. It will never get down to zero.” (Ed. note: his comments highlight real intellectual-property and privacy questions.)
A recent study found about 22% of computer science papers show AI content (research). OpenAI’s CFO Sarah Friar outlined plans for new pricing tied to outcomes and value creation in a blog post, saying licensing and outcome-based models will share in created value. Altman added that models are not yet ready for fully autonomous closed-loop research, calling truly independent operation still a reasonably long way off.
