ST-LLM: benchmark results
Provider: Other.
Aggregate Rating 1412 ± 42, rank #5548 of 10916 rated models, from 28 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| EgoMonth - Object Location | 58.2 | Accuracy (%; 206 object location questions; four-option mult | 63.6 |
| EgoMonth - Personality Inference | 56.2 | Accuracy (%; 16 personality inference questions; four-option | 45.5 |
| CG-Bench - Clue Grounding mIoU | 2.13 | Temporal Grounding mIoU (%) | 42.9 |
| CG-Bench - Accuracy@IoU | 1.27 | Accuracy@IoU (%) | 38.1 |
| EgoMonth - Event Time | 42.7 | Accuracy (%; 117 event time questions; four-option multiple- | 22.7 |
| EgoMonth - Object Counting | 31.6 | Accuracy (%; 57 object counting questions; four-option multi | 22.7 |
| EgoMonth - Procedure Planning | 46 | Accuracy (%; 126 procedure planning questions; four-option m | 22.7 |
| CG-Bench | 24.7 | Long-Video MCQ Accuracy (%) | 19 |
| CG-Bench - Clue MCQ | 39.1 | Clue-Interval MCQ Accuracy (%) | 19 |
| EgoMonth - Habit Inference | 65.2 | Accuracy (%; 138 habit inference questions; four-option mult | 18.2 |
| EgoMonth - Self-Localization | 35.2 | Accuracy (%; 85 self-localization questions; four-option mul | 18.2 |
| CG-Bench - Open-Ended | 20 | Open-Ended Accuracy (%) | 14.3 |
Interactive version: theaggregate.ai/model?slug=st-llm · How It Works · Data refreshed daily, snapshot 2026-11-08.