Llama 3.2 90B: benchmark results
Meta's 90B vision-language base model from Llama 3.2, adding cross-attention image adapters to Llama text weights with a 128K context. Provider: Meta. Released 2024-09-25. Access: Open.
Aggregate Rating 1556 ± 14, rank #596 of 1656 rated models, from 66 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| DermoBench - Diagnosis (25-Way Triage) | 51.65 | Accuracy (%; 25-way coarse-category MCQA, in-distribution) | 88.2 |
| VLRS-Bench - Evaluation Reasoning | 43 | Score (%) on the 250 post-event Evaluation Reasoning (single | 86.4 |
| VLRS-Bench - Planning Reasoning | 40.8 | Score (%) on the 250 pre-event Planning Reasoning (single im | 77.3 |
| Translation Set1 to en spBleu | 43.7 | Set1→en spBLEU (imported) | 75 |
| VLRS-Bench - Morphological Prediction | 35.2 | Score (%) on the 125 Spatiotemporal Morphological Prediction | 71.1 |
| AI for Education Pedagogy - Technology | 82.08 | Accuracy (%) | 67.8 |
| MMMU Benchmark | 60.3 | Validation Score | 65.8 |
| DermoBench - Reasoning | 50.38 | Reasoning Score (0-100; mean of two Gemini-2.5-Pro judge sco | 64.7 |
| VLRS-Bench - Causal Reasoning | 36.8 | Score (%) on the 125 Causal Reasoning (single image) questio | 63.6 |
| VLRS-Bench - Counterfactual Reasoning | 36.4 | Score (%) on the 125 Counterfactual Reasoning (single image) | 63.6 |
| MM-MT-Bench | 5.5 | Score (points) | 58.3 |
| DynaMath | 44 | Score (%) | 50 |
Interactive version: theaggregate.ai/model?slug=llama-3-2-90b · How It Works · Data refreshed daily, snapshot 2026-10-20.