whichllm — Find local LLMs for your hardware. Detects GPU/CPU/RAM 🖥️ Ranks models by fit 📊 Uses HuggingFace data 🔄 Outputs CLI/JSON/Python 🧩

Oh, this is such a nifty little tool. If you're looking to find out what LLMs you can run locally, use this. whichllm

ハードウェアの情報をみて、最適なローカル LLM を教えてくれる whichllmっていう OSS よさそう。 Codexとか Claude Code に投げるまでもない処理って意外とあるんですよね。 CS ログのカテゴライズとか gemma で十分でしたし。 単発で軽量で、並行してやりたい処理はローカル LLMとの相性いいので、適切な作業を寄せていきたいですね。

whichllm ranks local LLMs by actual benchmark scores, not just model size. Auto-detects your GPU/CPU/RAM and pulls live HuggingFace data to tell you what runs best on your hardware—then downloads and…

Show HN: Find the best local LLM for your hardware, ranked by benchmarks (

Just ran the `whichllm` CLI and the VRAM math is so good. It calculates exact limits for your GPU down to the GQA KV cache overhead and MoE active params. It literally refuses to hand you the biggest…

Picking a local model by VRAM fit alone is how you waste weekends, so I like whichllm ranking models by real benchmarks and speed for your actual hardware