True Index: Domain Rankings

Private Top-Tier (S & A) standing across 3 core domains
Agentic & Enterprise91.0p
#4
3 Top-Tier private tests
Coding & SWE88.8p
#5
2 Top-Tier private tests
Math & Scientific Reasoning68.6p
#14
2 Top-Tier private tests

Head-to-Head: Clean Evals vs. Legacy Benchmarks

3 matched domains · positive boost means higher legacy scores
Capability Domain Clean Evals (Tiers S & A) Legacy Benchmarks (Tiers C & D) Public Lift (Boost)
Agentic & Enterprise91.0p (3)79.3p (2)-11.7p
Coding & SWE88.8p (2)98.0p (1)+9.2p
Math & Scientific Reasoning68.6p (2)84.8p (2)+16.2p
Clean Evals Avg (Tiers S & A)
84.6p
10 private, leak-resistant tests
Legacy Benchmarks Avg (Tiers C & D)
87.4p
5 public or aging tests
Matched Public Lift (Boost)
+3.9p
Near zero or negative indicates clean generalization
Machine-Readable Export

Audited Benchmark Performance of Gemini 4 Carbon

Grouped by benchmark integrity so private capability signal stands apart from aging public tests.

Top-Tier Validated Tests (Tiers S & A) (10)

Private, leak-resistant tests that build real capability signal.

BenchmarkBenchmark Integrity StatusDomainRankPercentile
MKA (Multimodal Knowledge Agent)Private / FreshAgentic & Enterprise#3/2399.6p
AutomationBench-AAPrivate / FreshAgentic & Enterprise#2/5098.0p
Vibe Code Bench 1-100 v1.1Private / FreshCoding & SWE#2/1095.6p
FrontierMathPrivate / FreshMath & Scientific Reasoning#5/2789.9p
FrontierCode-AA 2.0Private / FreshCoding & SWE#10/4681.9p
AA-Briefcase v1.1Private / FreshAgentic & Enterprise#13/5075.5p
MLCR-AAPrivate / FreshMath & Scientific Reasoning#26/4647.2p
AA-Omniscience v1.1Private / FreshGeneral Knowledge & QA#3/4797.3p
LiveBenchPrivate / FreshGeneral Knowledge & QA#7/2887.8p
ARC-AGI-2Private / FreshGeneral Knowledge & QA#6/2285.1p

Mid-Tier Evaluations (Tier B) (2)

Partially exposed or transitional evaluations with moderate leakage risk.

BenchmarkBenchmark Integrity StatusDomainRankPercentile
HLE-DiamondSemi-ExposedGeneral Knowledge & QA#4/4993.8p
VIALS-AASemi-ExposedMath & Scientific Reasoning#3/2390.9p

Legacy & Vulnerable Benchmarks (Tiers C & D) (5)

Public or aging benchmarks prone to overfitting and training-data leaks.

BenchmarkBenchmark Integrity StatusDomainRankPercentile
Terminal-Bench 4.0Public / AgingCoding & SWE#2/5098.0p
Terminal-Bench-Science 0.1Public / AgingMath & Scientific Reasoning#5/4195.0p
GDP.pdfPublic / AgingAgentic & Enterprise#9/5083.7p
GDPval-AA v2.1Public / AgingAgentic & Enterprise#10/4375.0p
Autoform-HardPublic / AgingMath & Scientific Reasoning#4/874.5p