Llama 3.2 90B Vision Instruct Benchmark Results

Llama 3.2 90B Vision Instruct: benchmark results

Meta Llama 3.2 90B vision instruction-tuned checkpoint. Provider: Meta. Released 2024-09-25. Access: Open.

Aggregate Rating 1547 ± 8, rank #622 of 1656 rated models, from 661 benchmark results.

Benchmark	Score	Metric	Percentile
AMVICC - MMVP Individual Accuracy	91.88	Individual Accuracy (%)	100
AMVICC - MMVP Individual Accuracy - Orientation and Direction	90	Individual accuracy (%; the 30 MMVP questions on 15 image pa	100
AMVICC - MMVP Individual Accuracy - Positional and Relational Context	90.63	Individual accuracy (%; the 32 MMVP questions on 16 image pa	100
AMVICC - MMVP Individual Accuracy - Presence of Specific Features	97.14	Individual accuracy (%; the 70 MMVP questions on 35 image pa	100
AMVICC - MMVP Individual Accuracy - Structural and Physical Characteristics	96.67	Individual accuracy (%; the 30 MMVP questions on 15 image pa	100
AMVICC - MMVP Individual Accuracy - Viewpoint and Perspective	100	Individual accuracy (%; the 18 MMVP questions on 9 image pai	100
AMVICC - MMVP Pair Accuracy	84.18	Pair Accuracy (%)	100
AMVICC - MMVP Pair Accuracy - Orientation and Direction	80	Pair accuracy (%; the 15 MMVP image pairs in the Orientation	100
AMVICC - MMVP Pair Accuracy - Positional and Relational Context	81.25	Pair accuracy (%; the 16 MMVP image pairs in the Positional	100
AMVICC - MMVP Pair Accuracy - Presence of Specific Features	94.29	Pair accuracy (%; the 35 MMVP image pairs in the Presence of	100
AMVICC - MMVP Pair Accuracy - Structural and Physical Characteristics	93.33	Pair accuracy (%; the 15 MMVP image pairs in the Structural	100
AMVICC - MMVP Pair Accuracy - Viewpoint and Perspective	100	Pair accuracy (%; the 9 MMVP image pairs in the Viewpoint an	100

Interactive version: theaggregate.ai/model?slug=llama-3-2-90b-vision-instruct · How It Works · Data refreshed daily, snapshot 2026-10-20.