Perception-LM-8B: benchmark results
Provider: Other.
Aggregate Rating 1605 ± 47, rank #963 of 10916 rated models, from 17 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| MoHallBench - Action-Action Hallucination | 34.8 | Q-PairAcc (%; share of adversarial pairs in which the model | 100 |
| MoHallBench - Object-Action Hallucination | 36.8 | Q-PairAcc (%; share of adversarial pairs in which the model | 100 |
| MoHallBench - Scene-Action Hallucination | 63.1 | Q-PairAcc (%; share of adversarial pairs in which the model | 100 |
| MoHallBench - Scene-Action Hallucination (Multiple Choice) | 83.5 | Multiple-choice accuracy (%; four-option multiple choice (gr | 100 |
| MoHallBench - Action-Action Hallucination (Multiple Choice) | 80.8 | Multiple-choice accuracy (%; four-option multiple choice (gr | 94.4 |
| Flat-Pack Bench - Part Tracking | 44.75 | Track: recover the correspondence of shuffled part IDs betwe | 93.1 |
| MoHallBench - Semantic Similarity Hallucination | 38.1 | Q-PairAcc (%; share of adversarial pairs in which the model | 88.9 |
| MoHallBench - Semantic Similarity Hallucination (Multiple Choice) | 84.8 | Multiple-choice accuracy (%; four-option multiple choice (gr | 88.9 |
| MoHallBench - Sequential Inference Hallucination | 22.3 | Q-PairAcc (%; share of adversarial pairs in which the model | 88.9 |
| MoHallBench - Visual Similarity Hallucination | 34.5 | Q-PairAcc (%; share of adversarial pairs in which the model | 88.9 |
| MoHallBench - Object-Action Hallucination (Multiple Choice) | 73.6 | Multiple-choice accuracy (%; four-option multiple choice (gr | 77.8 |
| MoHallBench - Sequential Inference Hallucination Rate (Generative) | 53.7 | Hallucination rate (%; share of free-form descriptions of tr | 55.6 |
Interactive version: theaggregate.ai/model?slug=perception-lm-8b · How It Works · Data refreshed daily, snapshot 2026-11-08.