Epoch AI - Superglue: leaderboard

Suite of eight challenging NLU tasks including textual entailment, coreference resolution, word-sense disambiguation, and reading comprehension requiring genuine reasoning.

Metric: Score. Source: epoch.ai. 8 models tracked.

Top models

#ModelScoreOverall rank
1T5 (11B)88.9#1652
2text-davinci-00171.8#1512

Interactive version: theaggregate.ai/benchmark?slug=epoch-ai-superglue · How It Works · Data refreshed daily, snapshot 2026-10-20.