PLLaVA-34B: benchmark results
Provider: Other.
Aggregate Rating 1495 ± 35, rank #2433 of 10916 rated models, from 75 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| OpenVLM Video - MVBench - Egocentric Navigation | 39.5 | Accuracy (%) | 88.9 |
| OpenVLM Video - Video-MME (w/o subs) - Spatial Reasoning | 83.9 | Accuracy (%) | 83.3 |
| OpenVLM Video - MMBench-Video - Hallucination | 1 | GPT-4 Rating (0-3) | 80.7 |
| OpenVLM Video - MVBench - Episodic Reasoning | 60.5 | Accuracy (%) | 70.8 |
| OpenVLM Video - MLVU - Generation Score | 4.26 | Judge Score (0-10) | 70 |
| OpenVLM Video - MLVU - Anomaly Recognition | 72.5 | Accuracy (%) | 69.4 |
| OpenVLM Video - MVBench - Unexpected Action | 77.5 | Accuracy (%) | 63.9 |
| OpenVLM Video - MVBench - Action Antonym | 83.5 | Accuracy (%) | 62.5 |
| OpenVLM Video - MVBench - Counterfactual Inference | 58 | Accuracy (%) | 58.3 |
| LongVideoBench | 53.5 | Test Total Accuracy (%) | 55.9 |
| OpenVLM Video - MVBench - Moving Direction | 37.5 | Accuracy (%) | 55.6 |
| OpenVLM Video - MVBench - Object Existence | 70 | Accuracy (%) | 55.6 |
Interactive version: theaggregate.ai/model?slug=pllava-34b · How It Works · Data refreshed daily, snapshot 2026-11-08.