Biggest AI model movers
Between 2026-10-17 and 2026-10-18, 0 models climbed and 0 dropped on the meo benchmark. Rank is the relative signal — a model climbs only by passing others on the un-leaked meo benchmark, not when a re-grade lifts the whole board.
Frequently asked questions
Which AI models improved the most recently?
Movers are computed from two or more benchmark snapshots. The first snapshot is recorded; rank changes will be shown once a second run is synced.
How are model movers calculated?
Each benchmark run is snapshotted per model (rank, accuracy, Effective Value). A "mover" is the change in a model’s meo rank between the two most recent snapshots — positive means it climbed relative to the field, independent of board-wide re-grades.
Why don’t board-wide accuracy changes count as movers?
A re-grade that shifts every model’s accuracy by a similar amount barely changes the relative order, so it is not a "mover." Movers surface models that changed position relative to the rest of the field — the signal that actually matters for ranking.
Related: full leaderboard · changelog · methodology.