LLaVA-Next-LLaMA3-8B: benchmark results
Provider: LLaVA Team. Access: Open.
Aggregate Rating 1438 ± 26, rank #4289 of 10916 rated models, from 78 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| X-PCR - Lesion Characterization | 79.44 | Accuracy (%) at stage 3, characterizing lesion morphology an | 95 |
| MIR-SafetyBench - Temporal Continuity | 54.26 | Attack Success Rate (%; lower is safer) | 83.3 |
| MIR-SafetyBench - Temporal Jump | 52.48 | Attack Success Rate (%; lower is safer) | 83.3 |
| MIR-SafetyBench - Spatial Juxtaposition | 53.42 | Attack Success Rate (%; lower is safer) | 77.8 |
| MIR-SafetyBench | 60.87 | Attack Success Rate (%; lower is safer) | 72.2 |
| MIR-SafetyBench - Logical Analogy | 55.66 | Attack Success Rate (%; lower is safer) | 72.2 |
| MIR-SafetyBench - Logical Causality | 65.36 | Attack Success Rate (%; lower is safer) | 72.2 |
| MirrorBench - Level 2 (Implicit Mirror Discovery) | 0.11 | Average score (mean of task success rate, step-wise improvem | 70.6 |
| Epoch AI - Science Qa | 73.7 | Score | 69.6 |
| MIR-SafetyBench - Semantic Complementarity | 60.71 | Attack Success Rate (%; lower is safer) | 69.4 |
| MIR-SafetyBench - Spatial Embedding | 61.77 | Attack Success Rate (%; lower is safer) | 66.7 |
| MIR-SafetyBench - Semantic Relevance | 57.24 | Attack Success Rate (%; lower is safer) | 61.1 |
Interactive version: theaggregate.ai/model?slug=llava-next-llama3-8b · How It Works · Data refreshed daily, snapshot 2026-11-08.