True Index: Domain Rankings

Private Top-Tier (S & A) standing across 3 core domains
Agentic & Enterprise81.2p
#8
5 Top-Tier private tests
Coding & SWE79.0p
#9
5 Top-Tier private tests
Math & Scientific Reasoning82.6p
#6
3 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 & Enterprise81.2p (5)79.2p (2)-1.9p
Coding & SWE79.0p (5)75.1p (3)-3.9p
Math & Scientific Reasoning82.6p (3)94.1p (3)+11.5p
Clean Evals Avg (Tiers S & A)
78.3p
15 private, leak-resistant tests
Legacy Benchmarks Avg (Tiers C & D)
82.8p
8 public or aging tests
Matched Public Lift (Boost)
+1.6p
Near zero or negative indicates clean generalization
Machine-Readable Export

Audited Benchmark Performance of Claude Sonnet 5.5

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

Top-Tier Validated Tests (Tiers S & A) (15)

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

BenchmarkBenchmark Integrity StatusDomainRankPercentile
AA-Briefcase v1.1Private / FreshAgentic & Enterprise#1/50100.0p
MLCR-AAPrivate / FreshMath & Scientific Reasoning#2/4697.6p
MKA (Multimodal Knowledge Agent)Private / FreshAgentic & Enterprise#6/2387.4p
FrontierCode-AA 2.0Private / FreshCoding & SWE#8/4686.4p
Code2Video-AAPrivate / FreshCoding & SWE#2/2384.9p
AutomationBench-AAPrivate / FreshAgentic & Enterprise#10/5081.6p
FrontierMathPrivate / FreshMath & Scientific Reasoning#11/2770.9p
Harvey LAB-AA v1.1Private / FreshAgentic & Enterprise#16/2148.5p
Code Migration v1.1Private / FreshCoding & SWE#1/2496.6p
AA-3DPrivate / FreshCoding & SWE#4/2291.1p
AA-AnalystAgentPrivate / FreshAgentic & Enterprise#3/2488.3p
CritPt v2Private / FreshMath & Scientific Reasoning#6/4079.3p
AA-Omniscience v1.1Private / FreshGeneral Knowledge & QA#14/4773.2p
LiveBenchPrivate / FreshGeneral Knowledge & QA#13/2867.7p
CyberBench v1.1Private / FreshCoding & SWE#19/2536.0p

Mid-Tier Evaluations (Tier B) (2)

Partially exposed or transitional evaluations with moderate leakage risk.

BenchmarkBenchmark Integrity StatusDomainRankPercentile
HLE-DiamondSemi-ExposedGeneral Knowledge & QA#12/4977.1p
VIALS-AASemi-ExposedMath & Scientific Reasoning#8/2365.9p

Legacy & Vulnerable Benchmarks (Tiers C & D) (8)

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

BenchmarkBenchmark Integrity StatusDomainRankPercentile
BioMysteryBenchPublic / AgingMath & Scientific Reasoning#2/23100.0p
HealthBenchPublic / AgingMath & Scientific Reasoning#1/16100.0p
Terminal-Bench 4.0Public / AgingCoding & SWE#4/5092.9p
GDPval-AA v2.1Public / AgingAgentic & Enterprise#4/4389.1p
SWE Atlas - Codebase QnAPublic / AgingCoding & SWE#4/1588.1p
Terminal-Bench-Science 0.1Public / AgingMath & Scientific Reasoning#10/4182.4p
GDP.pdfPublic / AgingAgentic & Enterprise#16/5069.4p
SWE-bench Pro V2Public / AgingCoding & SWE#6/844.4p