← Ground Floor

Will-It-Run?

Local model fit and speed advisor for open-weight LLMs. Pick your Mac and a model — get an honest can-you-run-it verdict with required memory and expected tokens/sec, from Ground Floor’s own measured data. Runs entirely in your browser; nothing is sent anywhere.

Hardware RAM / VRAM Bandwidth GB/s Context 8192

I raised my GPU memory limit (advanced) open-weight only hide ❌

Model	Verdict	Quant	tok/s	GB	Max ctx	Reason

tok/s and KV-cache are ESTIMATES; bandwidth values marked unverified are approximate; verdicts assume a clean machine. Discrete GPU mode ignores CPU offload in v1.