Together Instant Clusters General Availability — Slurm-on-Kubernetes upgrade & DCI launch

AI Impact Summary

Together AI has announced General Availability for their Instant Clusters, introducing a new Slurm-on-Kubernetes stack with significant reliability improvements and in-place migration capabilities. This upgrade includes self-healing worker daemons, durable job accounting, and improved GPU state tracking, addressing previous operational challenges. The shift to prepaid billing and dynamic rate limits further streamlines cost management, and the launch of Dedicated Container Inference (DCI) provides users with greater control over model deployment.