PerceptionBench with tools

PerceptionBench fine-grained visual perception tasks run with tool use, as reported by Meta.

ModelsModels with a result.3
Top resultBest result on this benchmark.—
Top-3 spreadPoints from first to third.—
YearYear of release.2026

Results

3 results
#ModelResultThe score from the source.
—Muse Ember80.6%
—GPT-6 Luna62.8%