benchgap
Other · model

Phi-4 Multimodal Instruct benchmark scores

As of 2026-11-22, Phi-4 Multimodal Instruct (Other) has measured scores on 3 benchmarks and estimated scores on 21 more.

BenchmarkScoreSource
AA-GPQA Diamond31.5%measured
AA-HLE5.0%measured
Artificial Analysis Intelligence Index10.0%estimated ± 7.4 pp, medium confidence
CritPt0.2%estimated ± 3.4 pp, high confidence
GPQA Diamond69.4%estimated ± 2.4 pp, high confidence
HLE16.9%estimated ± 8.5 pp, medium confidence
AA-Omniscience Accuracy18.5%estimated ± 7.7 pp, low confidence
AA-MMMU-Pro14.5%measured
MMLU-Pro42.2%estimated ± 7.0 pp, medium confidence
GPQA37.6%estimated ± 6.1 pp, medium confidence
MMLU67.3%estimated ± 8.3 pp, medium confidence
Vals GPQA Diamond30.7%estimated ± 3.2 pp, medium confidence
Vals MMLU-Pro53.8%estimated ± 3.0 pp, medium confidence
MMMLU66.2%estimated ± 1.7 pp, medium confidence
AA MMLU-Pro85.7%estimated ± 1.3 pp, medium confidence
ARC-AGI-161.2%estimated ± 8.1 pp, low confidence
ARC-AGI-226.1%estimated ± 14.4 pp, low confidence
HLE w/o tools3.4%estimated ± 5.4 pp, low confidence
ARC-AGI-31.3%estimated ± 9.8 pp, low confidence
ARC-AGI-3 Provider Adapter2.7%estimated ± 7.3 pp, low confidence
C-Eval87.8%estimated ± 1.2 pp, low confidence
MMLU-Redux84.5%estimated ± 3.3 pp, medium confidence
MMMU-Pro18.8%estimated ± 1.5 pp, medium confidence
SimpleQA4.1%estimated ± 7.8 pp, low confidence