benchgap
Google · model

Gemini 4 Hydrogen benchmark scores

As of 2026-11-18, Gemini 4 Hydrogen (Google) has measured scores on 16 benchmarks and estimated scores on 70 more.

BenchmarkScoreSource
AA-GPQA Diamond90.9%estimated ± 11.5 pp, low confidence
AA-HLE53.7%estimated ± 9.2 pp, medium confidence
AA-SciCode62.1%estimated ± 2.5 pp, high confidence
Artificial Analysis Intelligence Index53.9%measured
CritPt28.7%estimated ± 5.3 pp, medium confidence
GDPval-AA59.2%estimated ± 1.7 pp, medium confidence
GPQA Diamond92.9%estimated ± 2.5 pp, high confidence
HLE52.9%estimated ± 10.3 pp, low confidence
AA-Omniscience Accuracy50.5%measured
Terminal-Bench 2.189.1%estimated ± 5.4 pp, low confidence
BrowseComp90.9%estimated ± 11.3 pp, low confidence
HLE w/ tools66.4%estimated ± 6.8 pp, low confidence
AA Coding Index83.1%estimated ± 1.2 pp, medium confidence
AA-MMMU-Pro86.6%estimated ± 3.0 pp, medium confidence
BioMysteryBench (human-difficult)46.8%estimated ± 7.8 pp, low confidence
AutomationBench50.6%estimated ± 6.1 pp, low confidence
JobBench72.5%estimated ± 11.3 pp, low confidence
SWE-bench Pro84.2%estimated ± 4.2 pp, high confidence
MMLU-Pro85.5%estimated ± 5.1 pp, low confidence
GPQA93.0%estimated ± 6.9 pp, medium confidence
MMLU90.5%estimated ± 0.9 pp, medium confidence
Vals GPQA Diamond92.7%estimated ± 5.7 pp, medium confidence
Vals LiveCodeBench91.5%estimated ± 3.0 pp, medium confidence
Vals MMLU-Pro88.4%estimated ± 2.9 pp, high confidence
AA MMLU-Pro88.8%estimated ± 0.8 pp, high confidence
AA-AnalystAgent53.9%estimated ± 2.7 pp, medium confidence
ApprenticeBench15.4%estimated ± 12.5 pp, low confidence
ARC-AGI-191.4%estimated ± 7.9 pp, low confidence
ARC-AGI-283.7%estimated ± 14.6 pp, low confidence
CursorBench 3.274.6%estimated ± 2.0 pp, medium confidence
Terminal-Bench 3.044.2%estimated ± 3.2 pp, medium confidence
FrontierCode 1.1 Main51.1%estimated ± 3.8 pp, low confidence
FrontierSWE v253.1%measured
MazeBench1.0%measured
ParseBench-173.0%estimated ± 5.7 pp, low confidence
Terminal-Bench 4.156.4%measured
Terminal-Bench Hard68.1%estimated ± 2.8 pp, low confidence
Vals SWE-bench100.0%estimated ± 3.7 pp, medium confidence
Terminal-Bench 2.1 (Vals AI)80.3%estimated ± 6.9 pp, low confidence
VulcanBench v387.6%estimated ± 3.9 pp, medium confidence
CWE-Bench-AA62.9%estimated ± 4.1 pp, medium confidence
DeepsecBench-AA31.8%estimated ± 6.1 pp, low confidence
Harvey LAB-AA v1.18.9%estimated ± 3.5 pp, medium confidence
HealthBench Professional (length-adjusted, Artificial Analysis)62.3%estimated ± 4.3 pp, medium confidence
HealthBench Professional (raw, Artificial Analysis)69.9%estimated ± 1.1 pp, medium confidence
Bug Hunt Bench32.7%measured
Chartography (no tools)66.6%measured
Chartography (tools)90.2%estimated ± 2.3 pp, medium confidence
CursorBench 4.054.3%measured
CWE-bench v157.3%estimated ± 5.3 pp, medium confidence
DeepSWE74.0%estimated ± 4.1 pp, medium confidence
DRACO1.7%estimated ± 0.7 pp, low confidence
FrontierCode 2.038.9%measured
HLE w/o tools59.0%estimated ± 5.0 pp, high confidence
OSWorld 2.178.6%estimated ± 6.1 pp, medium confidence
OSWorld 2.1 Offline77.6%measured
PostTrainBench v1.148.6%estimated ± 1.7 pp, medium confidence
SWE-bench Multimodal73.6%estimated ± 8.3 pp, low confidence
Terminal-Bench 4.058.3%estimated ± 3.7 pp, high confidence
Terminal-Bench-Science 0.151.4%measured
Toolathlon-Verified78.0%estimated ± 1.8 pp, low confidence
VulcanBench Frontier v492.2%estimated ± 1.8 pp, high confidence
AA AutomationBench76.1%measured
GDP.pdf27.6%measured
AA Terminal-Bench 4.053.5%measured
CWE-bench68.1%measured
MKA53.4%measured
Hermes Bench100.0%estimated ± 1.8 pp, low confidence
Hermes Index88.9%estimated ± 2.7 pp, low confidence
ParseBench-282.6%estimated ± 6.4 pp, low confidence
ExploitGym13.9%estimated ± 6.4 pp, medium confidence
MMLU-Redux97.3%estimated ± 0.8 pp, low confidence
NL2Repo64.6%estimated ± 6.5 pp, low confidence
SuperGPQA99.9%estimated ± 9.3 pp, low confidence
PostTrainBench37.5%estimated ± 2.4 pp, high confidence
Finance Agent v256.2%estimated ± 3.8 pp, medium confidence
FrontierCode 1.1 Extended62.1%estimated ± 3.0 pp, medium confidence
HealthBench (length-adjusted)60.4%estimated ± 3.7 pp, high confidence
HealthBench Professional62.3%estimated ± 4.2 pp, medium confidence
HLE-Verified46.9%estimated ± 7.1 pp, low confidence
ScarfBench69.3%estimated ± 4.5 pp, low confidence
STEMBench2.6%estimated ± 0.2 pp, medium confidence
Agents' Last Exam50.1%estimated ± 10.7 pp, low confidence
OpenHarmony Bench v1.263.8%estimated ± 11.0 pp, low confidence
SWE-Marathon51.2%estimated ± 2.1 pp, medium confidence
BioBasicsBench34.1%estimated ± 3.8 pp, medium confidence