What it runs

The capability library

Everything a 128GB box runs, verified the hard way: byte-exact sizes, hands-on speeds, licences read, provenance on every score.

The model library

Every model that fits, byte-verified — text, image, video, voice, music, plus what does not fit.

Who measured what, on which harness — the three score families kept separate.

OpenClaw, Hermes and the coding agents — every local claim verified against the tool's own docs.

ChatGPT, but it's yours — the front-ends and the servers underneath.

Capacity, not velocity — measured pair results and honest buy-later guidance.