Nebius(@nebiusai):New benchmark results from @weka and Nebius show the impact of shared KV caching on sustained agentic inference. Running on @nvidia HGX B300 instances in Nebius AI Cloud, WEKA’s Augmented Memory Grid served 2.4× more requests than HBM-only while keeping median time to first token near one second. Worth a read for anyone working through the infrastructure demands of long-context agents:

Nebius

@nebiusai

The ultimate cloud for AI innovators. For GenAI open-source model endpoints, check out @nebiustf.

参加 June 2023

928 フォロー中 42.6K ファン

Nebius@nebiusai

2026.09.24 18:29

New benchmark results from @weka and Nebius show the impact of shared KV caching on sustained agentic inference. Running on @nvidia HGX B300 instances in Nebius AI Cloud, WEKA’s Augmented Memory Grid served 2.4× more requests than HBM-only while keeping median time to first token near one second. Worth a read for anyone working through the infrastructure demands of long-context agents: