GLM-5.3 served locally on two DGX Stations
What happened
On November 14, 2026, a Hacker News front-page post described NVIDIA serving GLM-5.3 locally on two DGX Stations. The machines are linked by ConnectX-8 RDMA and run vLLM tensor parallelism at TP=2, serving the 753B-parameter GLM-5.3 NVFP4 model from a single endpoint. The post reports 607.8 tok/s throughput.
Written by AI from the coverage · updated 2 days ago
Coverage
Follow the reports to see the story from different sides.
- Hacker News front pageServing GLM-5.3 at 607.8 tok/s across two Nvidia DGX stations with vLLM
NVIDIA 用两台通过 ConnectX-8 RDMA 互联的 DGX Station,以 vLLM 张量并行(TP=2)在单端点本地服务 753B 参数的 GLM-5.3 NVFP4 模型,实现 607.8 tok/s 吞吐。
Heat over time
Heat now 3·Comparable peak 10(Nov 14 23:00)·Comparable change over 24 hours -50%
The trend only compares accounts observed without gaps, so its range may be smaller than the current heat. Hover or tap the chart for each hour; the left and right arrow keys step through it.