Head-to-Head: Clean Evals vs. Legacy Benchmarks
| Capability Domain | Clean Evals (Tiers S & A) | Legacy Benchmarks (Tiers C & D) | Public Lift (Boost) |
|---|---|---|---|
| Agentic & Enterprise | 60.2p | 50.2p | -10.0p |
| Coding & SWE | 30.1p | 26.5p | -3.6p |
| Math & Scientific Reasoning | 39.0p | 68.8p | +29.8p |
Non-comparable pre-release tests omitted from audited scoring (model released 2026-08-12): Terminal-Bench-Science 0.1 (2026-08-27).
Clean Evals Avg (Tiers S & A)
43.3p
9 private, leak-resistant tests
Legacy Benchmarks Avg (Tiers C & D)
48.5p
4 public or aging tests
Matched Public Lift (Boost)
+5.4p
Near zero or negative indicates clean generalization
Machine-Readable Export
Audited Benchmark Performance of Grok 4.6
Grouped by benchmark integrity so private capability signal stands apart from aging public tests.
Top-Tier Validated Tests (Tiers S & A) (9)
Private, leak-resistant tests that build real capability signal.
| Benchmark | Benchmark Integrity Status | Domain | Rank | Percentile |
|---|---|---|---|---|
| AutomationBench-AA | Private / Fresh | Agentic & Enterprise | #19/50 | 63.3p |
| AA-Briefcase v1.1 | Private / Fresh | Agentic & Enterprise | #22/50 | 57.1p |
| MLCR-AA | Private / Fresh | Math & Scientific Reasoning | #32/46 | 33.1p |
| FrontierMath | Private / Fresh | Math & Scientific Reasoning | #21/27 | 32.5p |
| FrontierCode-AA 2.0 | Private / Fresh | Coding & SWE | #33/46 | 30.1p |
| AA-Omniscience v1.1 | Private / Fresh | General Knowledge & QA | #20/47 | 59.9p |
| CritPt v2 | Private / Fresh | Math & Scientific Reasoning | #20/40 | 51.3p |
| ARC-AGI-2 | Private / Fresh | General Knowledge & QA | #16/22 | 37.8p |
| LiveBench | Private / Fresh | General Knowledge & QA | #22/28 | 34.3p |
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 | #32/49 | 35.4p |
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 |
|---|---|---|---|---|
| GPQA Diamond | Public / Aging | Math & Scientific Reasoning | #5/17 | 68.8p |
| GDPval-AA v2.1 | Public / Aging | Agentic & Enterprise | #15/43 | 66.7p |
| GDP.pdf | Public / Aging | Agentic & Enterprise | #33/50 | 33.7p |
| Terminal-Bench 4.0 | Public / Aging | Coding & SWE | #37/50 | 26.5p |