EgoGPT-7B: benchmark results

Provider: Other.

Aggregate Rating 1495 ± 51, rank #2386 of 10556 rated models, from 16 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
EgoActionBench - Semantic Motion20.3Motion agreement judged by the language-model judge (0-100; 41.7
EgoEsportsQA - Reasoning35.43Accuracy (%) on the reasoning-level EgoEsportsQA questions o37.5
EgoEsportsQA34.73Overall accuracy (%) on the 1,745 EgoEsportsQA multiple-choi25
EgoActionBench - Semantic Verb18.9Verb agreement judged by the language-model judge (0-100; al22.2
EgoEsportsQA - Perception34.18Accuracy (%) on the perception-level EgoEsportsQA questions 18.8
EgoActionBench15.4EgoActionScore (0-100; mean of Temporal Overall and Semantic16.7
EgoActionBench - EPIC-KITCHENS-100 subset16EgoActionScore (0-100; mean of Temporal Overall and Semantic16.7
EgoActionBench - EgoLife subset15.9EgoActionScore (0-100; mean of Temporal Overall and Semantic16.7
EgoActionBench - Temporal F116Micro-averaged action F1 from one-to-one matching (0-100; al16.7
EgoActionBench - Semantic Hand24.1Hand score (0-100; hand-label agreement of the prediction se13.9
EgoActionBench - Ego4D subset14.5EgoActionScore (0-100; mean of Temporal Overall and Semantic11.1
EgoActionBench - Semantic20.4Semantic Overall (0-100; applicable Hand, Verb, Objects, Too11.1

Interactive version: theaggregate.ai/model?slug=egogpt-7b · How It Works · Data refreshed daily, snapshot 2026-11-17.