benchmark-model-kernels
82 / 100 · Low risk
Fine to try.
Inspect Hugging Face decoder layers on meta tensors and plan or run per-rank BF16, FP8, and NVFP4 GEMM or fused-MoE microbenchmarks with the bundled scripts and a local FlashInfer checkout. Use when choosing a model, GPU, TP, EP, or M/token-concurrency sweep; deriving common fused QKV and gate/up shapes without loading checkpoint weights; or using FlashInfer benchmark utilities. Do not use for end-to-end server throughput or request latency.
What this skill can do
Runs commands on your computer
Downloads and installs other software
Install
Paste this into your terminal. It downloads exactly the version we scanned, not whatever is newest.
if [ -e ~/.agents/skills/'benchmark-model-kernels' ] || [ -L ~/.agents/skills/'benchmark-model-kernels' ]; then echo 'A skill folder named benchmark-model-kernels already exists — not replacing it' >&2; false; else d=$(mktemp -d) && git clone --quiet --filter=blob:none --no-checkout 'https://github.com/NVIDIA/Model-Optimizer' "$d" && git -C "$d" checkout --quiet '6713c5a679fab80fdd9b688df01845f545c307b1' -- 'plugins/modelopt/skills/benchmark-model-kernels' && mkdir -p ~/.agents/skills && cp -R "$d"/'plugins/modelopt/skills/benchmark-model-kernels' ~/.agents/skills/'benchmark-model-kernels' && rm -rf "$d"; fi
Codex reads ~/.agents/skills directly. For Claude Code, Cursor, Copilot, Gemini CLI, Antigravity or OpenCode, also link it. For Claude Code: ln -s ~/.agents/skills/'benchmark-model-kernels' ~/.claude/skills/'benchmark-model-kernels' (other agents: ~/.cursor/skills, ~/.copilot/skills, ~/.gemini/skills, ~/.gemini/config/skills, ~/.config/opencode/skills). Or install through the finder skill, which does both.
Badge for your README
[](https://zalavijaysinh.github.io/skilltrust/repos/NVIDIA/Model-Optimizer/)
Score history
| Date | Version | Score |
|---|---|---|
| 2026-11-10 | 6713c5a | 82 |