True Index: Domain Rankings

Private Top-Tier (S & A) standing across 3 core domains
Agentic & Enterprise41.6p
#32
4 Top-Tier private tests
Coding & SWE47.1p
#29
6 Top-Tier private tests
Math & Scientific Reasoning36.3p
#32
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 & Enterprise41.6p (4)32.8p (2)-8.8p
Coding & SWE47.1p (6)66.3p (2)+19.3p
Math & Scientific Reasoning36.3p (3)56.3p (1)+20.0p
Non-comparable pre-release tests omitted from audited scoring (model released 2026-08-18): Terminal-Bench-Science 0.1 (2026-08-27), SWE-bench Pro V2 (2026-09-22).
Clean Evals Avg (Tiers S & A)
40.8p
14 private, leak-resistant tests
Legacy Benchmarks Avg (Tiers C & D)
51.8p
5 public or aging tests
Matched Public Lift (Boost)
+6.0p
Near zero or negative indicates clean generalization
Machine-Readable Export

Audited Benchmark Performance of GLM-5.3

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

Top-Tier Validated Tests (Tiers S & A) (14)

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

BenchmarkBenchmark Integrity StatusDomainRankPercentile
Vibe Code Bench 1-100 v1.1Private / FreshCoding & SWE#5/1068.0p
AutomationBench-AAPrivate / FreshAgentic & Enterprise#25/5051.0p
Code2Video-AAPrivate / FreshCoding & SWE#12/2348.2p
AA-Briefcase v1.1Private / FreshAgentic & Enterprise#27/5046.9p
MLCR-AAPrivate / FreshMath & Scientific Reasoning#32/4634.6p
FrontierCode-AA 2.0Private / FreshCoding & SWE#31/4634.5p
Harvey LAB-AA v1.1Private / FreshAgentic & Enterprise#20/2133.4p
FrontierMathPrivate / FreshMath & Scientific Reasoning#23/2733.0p
Code Migration v1.1Private / FreshCoding & SWE#11/2456.2p
CyberBench v1.1Private / FreshCoding & SWE#13/2553.9p
CritPt v2Private / FreshMath & Scientific Reasoning#22/4041.4p
AA-Omniscience v1.1Private / FreshGeneral Knowledge & QA#30/4738.2p
AA-AnalystAgentPrivate / FreshAgentic & Enterprise#18/2435.1p
AA-3DPrivate / FreshCoding & SWE#22/2221.5p

Mid-Tier Evaluations (Tier B) (1)

Partially exposed or transitional evaluations with moderate leakage risk.

BenchmarkBenchmark Integrity StatusDomainRankPercentile
HLE-DiamondSemi-ExposedGeneral Knowledge & QA#38/4922.9p

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

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

BenchmarkBenchmark Integrity StatusDomainRankPercentile
SWE Atlas - Codebase QnAPublic / AgingCoding & SWE#7/1569.4p
Terminal-Bench 4.0Public / AgingCoding & SWE#19/5063.3p
GDPval-AA v2.1Public / AgingAgentic & Enterprise#17/4358.5p
HealthBenchPublic / AgingMath & Scientific Reasoning#11/1656.3p
GDP.pdfPublic / AgingAgentic & Enterprise#46/507.1p