News summaryHPCwireOct 29, 2026, 06:28 PM
Rival.io Launches GraphX, Running DeepSeek V4 Flash on Just 8 NVIDIA L4 GPUs
Rival.io announced the launch of GraphX, its kernel-optimization and silicon-placement technology. [[fact:fact-w94h6aidh]] Rival.io also announced a deployment of DeepSeek V4 Flash on eight NVIDIA L4 GPUs. [[fact:fact-tg60vfs8e]] The eight-GPU configuration combines 192 GB of GPU memory and supports the model's context window and associated KV cache. [[fact:fact-rujw6hqvr]] Rival.io says GraphX can reduce open-weight model hosting costs by up to 95%, depending on the model and workload, versus equivalent internal or hyperscale-cloud deployments. [[fact:fact-jyjegcce1]]