$ omm contribute --no-color

Disk preflight passed: 50.2 GiB free on the tightest model volume. Each candidate is checked
again before download.
omm contribute - before you start:
  - Downloads, benchmarks, and deletes GGUF models repeatedly until you press Esc
  - Uses real bandwidth, disk space, and compute; runs unattended (no per-model confirmation)
  - Uploads every benchmark result per your current upload policy (ask)
  - Reserves space per candidate (central GGUF + worst-case engine copy + headroom); skips
anything that won't fit
  - Uses a fixed 1024-token context and 128-token batch for comparable results
  - Gates committed runtime memory before download; monitors paging and measurement stability
while running
  - Defers transient memory shortages up to three times instead of losing the candidate
  - Each benchmark has a 10-minute cutoff, with a status line every 30s
? Start contributing compute now? No
Cancelled.