Together GPU Clusters Adds Passive Health Checks, Node Repair, and OIDC Support

Together AI has introduced a set of new features to enhance the reliability of production-grade GPU clusters. Passive health checks continuously monitor node status without active probing, and with automatic node repair, failures can be quickly recovered. Additionally, Slurm job scheduler reliability has been strengthened to reduce task interruptions caused by scheduler issues. New OIDC support allows cluster authentication to integrate with existing enterprise single sign-on systems, improving security and management convenience. Startup scripts enable users to automatically execute custom configurations during node initialization, simplifying environment deployment. For teams using Together GPU clusters for large-scale model training or inference, these improvements mean less operational burden and higher cluster availability. Passive health checks and auto-repair reduce the need for manual inspections, Slurm enhancements improve job stability, and OIDC facilitates direct integration for enterprises already using unified identity authentication. Startup scripts standardize environment configuration, suitable for scenarios requiring frequent scaling.