DDN Infinia joins NetApp and VAST Data at Dutch neocloud Nebul
Dutch neocloud Nebul is using DDN's Infinia object-based OS for KV cache during inference. On the same eight GPUs, time to first token improved nearly 10x and cost per million input tokens fell almost 60%. Infinia joins NetApp ONTAP and VAST Data in Nebul's storage stack.
- Time to first token improved nearly 10x on the same eight GPUs
- Token response up to 5x faster, 2.5x more input tokens per second
- Cost per million input tokens down nearly 60%
- Nebul also uses NetApp ONTAP and VAST Data
Read next
AI