Benchmarks

Acceptance benchmarks by route

Inspect exact-model outcomes inside one job class without turning sparse samples into a score.

What is measured

Accepted terminal outcomes divided by all terminal outcomes for the exact model and job class shown.

Selection period

Trailing 7 complete UTC days, ending 2 September 2026.

Freshness

No complete data bucket has public evidence yet.

What this does not establish

This measures adoption or acceptance, never quality in general.

Not enough public evidence to rank an exact model route yet.