True Index: Domain Rankings

Private Top-Tier (S & A) standing across 3 core domains
Agentic & Enterprise92.8p
#2
5 Top-Tier private tests
Coding & SWE99.3p
#1
3 Top-Tier private tests
Math & Scientific Reasoning97.0p
#1
2 Top-Tier private tests

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

4 matched domains · positive boost means higher legacy scores
Capability Domain Clean Evals (Tiers S & A) Legacy Benchmarks (Tiers C & D) Public Lift (Boost)
Agentic & Enterprise92.8p (5)87.8p (1)-5.0p
Coding & SWE99.3p (3)78.0p (2)-21.3p
General Knowledge & QA98.4p (3)91.4p (1)-7.0p
Math & Scientific Reasoning97.0p (2)84.8p (4)-12.2p
Clean Evals Avg (Tiers S & A)
96.9p
13 private, leak-resistant tests
Legacy Benchmarks Avg (Tiers C & D)
85.5p
8 public or aging tests
Matched Public Lift (Boost)
-2.5p
Near zero or negative indicates clean generalization
Machine-Readable Export

Audited Benchmark Performance of GPT-6.2 Astra

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

Top-Tier Validated Tests (Tiers S & A) (13)

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

BenchmarkBenchmark Integrity StatusDomainRankPercentile
Harvey LAB-AA v1.1Private / FreshAgentic & Enterprise#1/21100.0p
Vibe Code Bench 1-100 v1.1Private / FreshCoding & SWE#1/10100.0p
FrontierMathPrivate / FreshMath & Scientific Reasoning#1/2799.5p
FrontierCode-AA 2.0Private / FreshCoding & SWE#3/4697.8p
MLCR-AAPrivate / FreshMath & Scientific Reasoning#4/4694.4p
AutomationBench-AAPrivate / FreshAgentic & Enterprise#4/5093.9p
MKA (Multimodal Knowledge Agent)Private / FreshAgentic & Enterprise#5/2391.5p
AA-Briefcase v1.1Private / FreshAgentic & Enterprise#10/5081.6p
AA-3DPrivate / FreshCoding & SWE#1/22100.0p
ARC-AGI-2Private / FreshGeneral Knowledge & QA#1/22100.0p
LiveBenchPrivate / FreshGeneral Knowledge & QA#2/28100.0p
AA-AnalystAgentPrivate / FreshAgentic & Enterprise#1/2497.2p
AA-Omniscience v1.1Private / FreshGeneral Knowledge & QA#4/4795.1p

Mid-Tier Evaluations (Tier B) (2)

Partially exposed or transitional evaluations with moderate leakage risk.

BenchmarkBenchmark Integrity StatusDomainRankPercentile
HLE-DiamondSemi-ExposedGeneral Knowledge & QA#2/4997.9p
VIALS-AASemi-ExposedMath & Scientific Reasoning#2/2395.5p

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

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

BenchmarkBenchmark Integrity StatusDomainRankPercentile
Autoform-HardPublic / AgingMath & Scientific Reasoning#1/8100.0p
Terminal-Bench-Science 0.1Public / AgingMath & Scientific Reasoning#1/41100.0p
BioMysteryBenchPublic / AgingMath & Scientific Reasoning#3/2396.6p
Terminal-Bench 4.0Public / AgingCoding & SWE#3/5095.9p
GDP.pdfPublic / AgingAgentic & Enterprise#7/5087.8p
SWE Atlas - Codebase QnAPublic / AgingCoding & SWE#9/1560.0p
HealthBenchPublic / AgingMath & Scientific Reasoning#14/1642.4p
MMMU-ProPublic / AgingGeneral Knowledge & QA#1/2291.4p