Source: https://github.com/gsma-labs/leaderboard/pull/50
Submission commit: 5930139dc9f364410eee8736f4ebf02d6eac6a9a
Validated all seven Inspect trajectories: full sample counts, no errors, per-sample accuracy reproduced, and submitted card agrees with logs.
Preserves benchmark scores and averages for all 90 existing models; updates ranks for the new entry. Refreshes JSON/CSV exports from the canonical dataset, including five existing entries missing from those exports.