Add Bekko System One v0 (17M, 68M, 400M) to the fine-tuned table, scored by us

Leaderboard: zero-shot models only; add self-reported entries; second table for models fitted on train

Restore the teacher description (4B-class); keep the request-shape note

Correct the teacher description; state the request shape

Add Liquid AI d1 (0.742)

Streamline the card: one leaderboard table, a chart, shorter notes

Tag meraGPT links with UTM parameters so card-driven visits can be attributed

Add Jeff (0.8B, 2B, Gemma4-E2B) to the leaderboard, scored 2026-09-29 through the Jev client path

Leaderboard: add Featherless Simple Jev (Qwen3.6-35B-A3B classifier), 0.716 acc, measured on the 400-case test split

Latency footnote: say exactly how it was measured

Leaderboard: Decider 1 beats Jev on accuracy, latency and price

Leaderboard: meraGPT Decider 1 is state of the art (0.768 accuracy, KL 0.096)

Replace the estimated System One row with a measured Jev 1.13.0 result

card: point usage examples at the canonical org path

card: explain each reference row, add ceilings and a saturation estimate

card: correct the prior baseline, which had been fitted on the benchmark

card: add Adaptive Classifier specialist baselines (MiniLM and ModernBERT)

rewrite dataset card

full dataset: 400-case benchmark (test) + 1200-case training split, verified disjoint

card: specialist vs generalist evaluation modes

benchmark-only: single test split of 400; add measured score reference points

card: remove personal attribution from citation

typed-decisions: 400 cases, 2000 typed decisions across 4 workflows