benchgap
Google · model

Gemini 2.5 Flash benchmark scores

As of 2026-11-22, Gemini 2.5 Flash (Google) has measured scores on 11 benchmarks and estimated scores on 59 more.

BenchmarkScoreSource
AA-GPQA Diamond68.3%measured
AA-HLE4.7%measured
Artificial Analysis Intelligence Index9.9%measured
CritPt1.4%measured
GDPval-AA0.0%estimated ± 11.0 pp, low confidence
GPQA Diamond69.2%estimated ± 2.4 pp, medium confidence
HLE16.8%estimated ± 8.5 pp, medium confidence
IFBench54.1%estimated ± 6.6 pp, low confidence
AA-LCR49.9%measured
AA-Omniscience Accuracy26.1%measured
BrowseComp20.3%estimated ± 13.0 pp, low confidence
IFEval88.5%estimated ± 2.8 pp, high confidence
LongBench v252.2%estimated ± 5.5 pp, medium confidence
τ²-bench results14.9%measured
AA-MMMU-Pro65.5%measured
BioMysteryBench (human-difficult)20.9%estimated ± 6.8 pp, low confidence
AA Agentic Index0.0%estimated ± 12.7 pp, low confidence
MMLU-Pro85.1%estimated ± 5.0 pp, medium confidence
AA-IFBench39.0%measured
FailureSensorIQ65.5%measured
GPQA70.8%estimated ± 6.1 pp, medium confidence
MMLU90.5%estimated ± 0.9 pp, high confidence
Vals GPQA Diamond67.4%estimated ± 3.2 pp, high confidence
Vals MMLU-Pro83.2%estimated ± 2.9 pp, high confidence
MMMLU86.8%estimated ± 0.7 pp, medium confidence
MMMU77.0%estimated ± 2.7 pp, high confidence
AA MMLU-Pro85.0%estimated ± 0.8 pp, high confidence
ARC-AGI-175.8%estimated ± 7.8 pp, low confidence
ARC-AGI-233.2%estimated ± 14.4 pp, low confidence
Mystery Game Puzzles1.0%estimated ± 4.7 pp, low confidence
OfficeQA Pro50.3%estimated ± 6.8 pp, low confidence
ParseBench-149.9%estimated ± 6.2 pp, medium confidence
HealthBench Professional (length-adjusted, Artificial Analysis)48.8%estimated ± 4.3 pp, low confidence
HealthBench Professional (raw, Artificial Analysis)41.3%estimated ± 1.1 pp, low confidence
HLE w/o tools23.7%estimated ± 4.9 pp, high confidence
ARC-AGI-31.3%estimated ± 9.8 pp, low confidence
ARC-AGI-3 Provider Adapter2.7%estimated ± 7.3 pp, low confidence
BinaryAudit26.7%estimated ± 8.9 pp, low confidence
CompileBench40.0%measured
OTelBench22.4%estimated ± 7.7 pp, low confidence
ParseBench-252.4%estimated ± 6.2 pp, medium confidence
CharXiv74.5%estimated ± 6.3 pp, low confidence
CharXiv w/o tools73.8%estimated ± 2.5 pp, medium confidence
AI2D_TEST88.0%estimated ± 2.0 pp, medium confidence
C-Eval88.5%estimated ± 1.0 pp, medium confidence
CountBench88.2%estimated ± 2.9 pp, low confidence
DynaMath6.5%estimated ± 3.0 pp, low confidence
ERQA41.3%estimated ± 3.1 pp, medium confidence
MathVision69.3%estimated ± 3.1 pp, medium confidence
MMLU-Redux94.3%estimated ± 0.8 pp, medium confidence
MMMU-Pro66.9%estimated ± 1.5 pp, high confidence
RealWorldQA76.5%estimated ± 2.9 pp, low confidence
ScreenSpot Pro68.2%estimated ± 10.8 pp, low confidence
SuperGPQA66.2%estimated ± 9.3 pp, medium confidence
V*1.3%estimated ± 7.5 pp, low confidence
Video-MME (w/o subtitle)77.2%estimated ± 1.7 pp, medium confidence
Video-MME (with subtitle)78.5%estimated ± 3.7 pp, low confidence
HealthBench Hard12.3%estimated ± 8.6 pp, low confidence
MedXpertQA (MM)20.1%estimated ± 6.5 pp, low confidence
MMLU-Pro (Arcee)82.4%estimated ± 4.0 pp, medium confidence
BioMysteryBench (human-solvable)81.7%estimated ± 0.6 pp, low confidence
HealthBench (length-adjusted)54.1%estimated ± 3.7 pp, medium confidence
HealthBench Professional56.7%estimated ± 4.2 pp, low confidence
HLE-Verified20.8%estimated ± 7.1 pp, low confidence
STEMBench2.4%estimated ± 0.2 pp, medium confidence
SimpleQA26.5%estimated ± 4.3 pp, medium confidence
MRCRv222.9%estimated ± 11.3 pp, low confidence
MMLongBench-Doc59.4%estimated ± 2.6 pp, low confidence
BioBasicsBench34.1%estimated ± 3.8 pp, low confidence
MMMU-Pro w/ Python69.7%estimated ± 2.0 pp, high confidence