Together AI Dedicated Inference - Per-GPU-hour single-tenant pricing

Fetched 2026-10-04. Dedicated Inference section: single-tenant GPU instances with guaranteed performance, support for custom models, autoscaling. All prices per GPU per hour. On-demand: HGX H100 $5.49, HGX B200 $8.99. H200, B300, GB200 NVL72, GB300 NVL72 all 'Contact us'. Reserved tiers all 'Contact sales' (/monthly-reserved). This is distinct from GPU Clusters (which has preemptible/reserved tiers published) and from serverless (per-token). The dedicated H100 at $5.49/GPU-hr sits between the GPU Clusters on-demand rate ($3.99) and the CoreWeave on-demand rate ($6.16). The dedicated B200 at $8.99 is the only published B200 rate from Together.