BLIP-2-FLAN-T5-XXL: benchmark results

Provider: Other.

Aggregate Rating 1338 ± 45, rank #7946 of 10489 rated models, from 26 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
CoreCognition Image only Spatiality63.64Accuracy (%)78.9
CoreCognition Image only Mechanicalreasoning55.11Accuracy (%)77
CoreCognition Image only Hierarchy60.52Accuracy (%)75.7
ChartBench - Chart Recognition60.05Acc+ (%)70
CoreCognition Image only Intentionality68.49Accuracy (%)60.5
CoreCognition Image only Continuity58.98Accuracy (%)52.6
CoreCognition Image only Boundary70.6Accuracy (%)48
CoreCognition Image only Toolusing91.81Accuracy (%)47.4
CoreCognition Image only Perspective33.73Accuracy (%)44.7
ChartBench16.7Five-task average, Acc+ and NumberQA (%)40
CoreCognition Image only Intuitivephysics50.86Accuracy (%)38.2
SpatiaLQA - Content F151.3Step-content F1 (%, 0-100): harmonic mean of the pooled reca37.5

Interactive version: theaggregate.ai/model?slug=blip-2-flan-t5-xxl · How It Works · Data refreshed daily, snapshot 2026-11-20.