Living proof · MultiLife

The same app, built twice to compare.

One brief, the same eleven phases, run end to end twice: once on a local Qwen, once on cloud open-weight models. Both shipped a working build. These are those builds, loaded live and unedited. Switch between them and play.

Run to start· switch Conquest / Classic· pick 2–4 teams· let it stabilize for the endgame verdict

Same PRD. Same eleven phases. Two model stacks. Nothing here is pre-recorded or hand-finished, it's exactly what each run produced. The cloud build cost $6.90 and finished faster; the local build cost nothing but power and ran clean.