SAIR soft-launches a collaborative AI workspace for mathematics
Open Math OS puts existing open models and mathematical computing tools into a shared research workspace; its creators describe the launch as an early release.
A first workspace for the Open Math Model initiative
SAIR has soft-launched Open Math OS, a shared workspace intended to help mathematicians use AI in research. In an October 16 announcement, mathematician Terence Tao described the release as an early version built around existing open models, following the initiative announced in September.
What researchers can use
Tao names DeepSeek V4.1 Flash, Kimi K3 and GLM 5.3 Flash among the models hosted at launch. He suggests uses such as discussing or summarizing mathematical arguments and formalizing medium-sized problems in Lean, while cautioning that these models are behind leading frontier systems and are not suited to solving difficult problems in a single attempt. These are the project’s assessments, not independently verified performance results.
The service emphasizes team accounts, where collaborators can see one another’s prompts. Its website also describes shared projects, scientific computing environments and tools including Lean, SageMath and SymPy. SAIR says project data is private by default and that using work for training requires explicit consent.
Access and limits
Tao says the service runs on donated computing resources and that free accounts are being granted to academic applicants, with availability potentially changing as demand and donations vary. The project is still seeking partners to develop a more mathematics-focused model. Researchers should check generated arguments and formal statements carefully: a shared workspace and access to proof tools do not by themselves establish the correctness of a result.