noonghunna

club-3090. Community recipes for serving LLMs on RTX 3090/4090/5090 CUDA gpus. Multi-engine (vLLM, llama.cpp, ik_llama) and model-agnostic. Currently shipping Qwen3.6-27B Qwen3.6 35B Gemma 4 26B Gemma 4 31B configs for 1× and 2× cards.

qwen36-27b-single-3090. Shell

qwen36-dual-3090. Qwen3.6-27B on dual RTX 3090 — TP=2 recipe, vLLM nightly, MTP + fp8 KV, validated for concurrent serving

benchlocal-cli. CLI port of BenchLocal quality bench packs — runs LLM behavioral evals against any OpenAI-compatible endpoint. Companion to club-3090.