ci_leaderboard recipe: judge-free, 30 rows/axis, small-n directional scores.
| # | model | Elo (BT-MLE) | MS (public) | MS (experimental ⚠) | knowledge | nlu | axes | run date |
|---|---|---|---|---|---|---|---|---|
| 1 | qwen2.5-0.5b-instruct | 1053.3 | 0.4000 [0.28,0.53] | n/a | 0.567 n=30 | 0.233 n=30 | 2 | 2026-07-31 |
| 2 | qwen2.5-1.5b-instruct | 1020.3 | 0.3833 [0.25,0.50] | n/a | 0.533 n=30 | 0.233 n=30 | 2 | 2026-07-31 |
| 3 | smollm2-1.7b-instruct | 926.4 | 0.3167 [0.22,0.42] | n/a | 0.500 n=30 | 0.133 n=30 | 2 | 2026-07-31 |
90 verifiable pairwise matches fed this board. Elo ranking is BT-MLE (order-invariant); a model is credibly above another only when their bootstrap CIs don't overlap — see glossobench elo for full CIs. This page is generated by CI from a small, judge-free, tiny-row-cap recipe (directional / demo only, not a real GlossoBench-5/7 claim) — see glossobench for the real harness.