True Index: Domain Rankings
Agentic & Enterprise78.7p
#10
4 Top-Tier private testsCoding & SWE54.8p
#23
1 Top-Tier private testMath & Scientific Reasoning47.2p
#23
1 Top-Tier private testHead-to-Head: Clean Evals vs. Legacy Benchmarks
| Capability Domain | Clean Evals (Tiers S & A) | Legacy Benchmarks (Tiers C & D) | Public Lift (Boost) |
|---|---|---|---|
| Agentic & Enterprise | 78.7p | 79.0p | +0.2p |
| Coding & SWE | 54.8p | 67.3p | +12.5p |
| Math & Scientific Reasoning | 47.2p | 62.3p | +15.1p |
Clean Evals Avg (Tiers S & A)
61.5p
8 private, leak-resistant tests
Legacy Benchmarks Avg (Tiers C & D)
69.5p
4 public or aging tests
Matched Public Lift (Boost)
+8.8p
Near zero or negative indicates clean generalization
Machine-Readable Export
Audited Benchmark Performance of Qwen4 Max
Grouped by benchmark integrity so private capability signal stands apart from aging public tests.
Top-Tier Validated Tests (Tiers S & A) (8)
Private, leak-resistant tests that build real capability signal.
| Benchmark | Benchmark Integrity Status | Domain | Rank | Percentile |
|---|---|---|---|---|
| AA-Briefcase v1.1 | Private / Fresh | Agentic & Enterprise | #6/50 | 89.8p |
| Harvey LAB-AA v1.1 | Private / Fresh | Agentic & Enterprise | #5/21 | 88.2p |
| AutomationBench-AA | Private / Fresh | Agentic & Enterprise | #8/50 | 85.7p |
| FrontierCode-AA 2.0 | Private / Fresh | Coding & SWE | #22/46 | 54.8p |
| MLCR-AA | Private / Fresh | Math & Scientific Reasoning | #26/46 | 47.2p |
| AA-Omniscience v1.1 | Private / Fresh | General Knowledge & QA | #17/47 | 66.6p |
| LiveBench | Private / Fresh | General Knowledge & QA | #14/28 | 64.3p |
| AA-AnalystAgent | Private / Fresh | Agentic & Enterprise | #13/24 | 51.1p |
Mid-Tier Evaluations (Tier B) (1)
Partially exposed or transitional evaluations with moderate leakage risk.
| Benchmark | Benchmark Integrity Status | Domain | Rank | Percentile |
|---|---|---|---|---|
| HLE-Diamond | Semi-Exposed | General Knowledge & QA | #16/49 | 68.8p |
Legacy & Vulnerable Benchmarks (Tiers C & D) (4)
Public or aging benchmarks prone to overfitting and training-data leaks.
| Benchmark | Benchmark Integrity Status | Domain | Rank | Percentile |
|---|---|---|---|---|
| GDPval-AA v2.1 | Public / Aging | Agentic & Enterprise | #6/43 | 84.4p |
| GDP.pdf | Public / Aging | Agentic & Enterprise | #13/50 | 73.5p |
| Terminal-Bench 4.0 | Public / Aging | Coding & SWE | #17/50 | 67.3p |
| Terminal-Bench-Science 0.1 | Public / Aging | Math & Scientific Reasoning | #18/41 | 62.3p |