# GPT-6 Luna - BenchAudit True Index Profile

**Integrity Profile: Clean Generalization**

> The Verdict: Performs Better on Fresh Private Evals Than Legacy Tests. GPT-6 Luna (evaluated at max effort) outperforms its legacy benchmark standing (34.1p) when moved to private, leak-resistant evaluations (40.9p), posting a -10.5p Low-Tier Boost across 4 matched domains, particularly in Coding & SWE (-26.8p boost) and Math & Scientific Reasoning (-17.5p boost)—a strong indicator of genuine generalization.

- **Canonical URL:** https://benchaudit.org/models/gpt-6-luna__max/
- **True Index Rank:** #28 of 44
- **Creator:** OpenAI
- **Weights:** Closed / API
- **Clean Evals Avg (Tiers S & A):** 40.9p (16 evals)
- **Legacy Benchmarks Avg (Tiers C & D):** 34.1p (8 evals)
- **Low-Tier Boost (Matched Delta):** -10.5p
- **Updated:** 2026-11-22

## True Index: Domain Rankings

| Capability Domain | Sub-Rank | Clean Evals (S/A) Avg | Top-Tier Evals |
| --- | ---: | ---: | ---: |
| Agentic & Enterprise | #35 | 34.3p | 5 |
| Coding & SWE | #25 | 50.3p | 5 |
| Math & Scientific Reasoning | #20 | 57.1p | 3 |

## 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 | 5 (34.3p) | 2 (37.8p) | +3.5p |
| Coding & SWE | 5 (50.3p) | 2 (23.5p) | -26.8p |
| General Knowledge & QA | 3 (22.0p) | 1 (35.5p) | +13.5p |
| Math & Scientific Reasoning | 3 (57.1p) | 3 (39.7p) | -17.5p |

## Audited Benchmark Performance of GPT-6 Luna

| Benchmark | Benchmark Integrity Status | Domain | Rank | Percentile |
| --- | --- | --- | ---: | ---: |
| [CyberBench v1.1](https://benchaudit.org/benchmarks/cyberbench/) | Private / Fresh | Coding & SWE | #4/25 | 80.7p |
| [MLCR-AA](https://benchaudit.org/benchmarks/mlcr-aa/) | Private / Fresh | Math & Scientific Reasoning | #16/46 | 69.3p |
| [HealthBench](https://benchaudit.org/benchmarks/healthbench/) | Public / Aging | Math & Scientific Reasoning | #10/16 | 61.0p |
| [FrontierMath](https://benchaudit.org/benchmarks/frontiermath/) | Private / Fresh | Math & Scientific Reasoning | #15/27 | 58.4p |
| [GDP.pdf](https://benchaudit.org/benchmarks/gdp-pdf/) | Public / Aging | Agentic & Enterprise | #21/50 | 57.1p |
| [AA-3D](https://benchaudit.org/benchmarks/aa-3d/) | Private / Fresh | Coding & SWE | #14/22 | 52.4p |
| [Harvey LAB-AA v1.1](https://benchaudit.org/benchmarks/harvey-lab-aa-v1-1/) | Private / Fresh | Agentic & Enterprise | #15/21 | 52.3p |
| [FrontierCode-AA 2.0](https://benchaudit.org/benchmarks/frontiercode-aa-2-0/) | Private / Fresh | Coding & SWE | #26/46 | 45.8p |
| [Code2Video-AA](https://benchaudit.org/benchmarks/code2video-aa/) | Private / Fresh | Coding & SWE | #13/23 | 44.6p |
| [CritPt v2](https://benchaudit.org/benchmarks/critpt-v2/) | Private / Fresh | Math & Scientific Reasoning | #21/40 | 43.7p |
| [AA-AnalystAgent](https://benchaudit.org/benchmarks/aa-analystagent/) | Private / Fresh | Agentic & Enterprise | #17/24 | 40.4p |
| [MKA (Multimodal Knowledge Agent)](https://benchaudit.org/benchmarks/mka/) | Private / Fresh | Agentic & Enterprise | #18/23 | 38.2p |
| [HLE-Diamond](https://benchaudit.org/benchmarks/hle-diamond/) | Semi-Exposed | General Knowledge & QA | #31/49 | 37.5p |
| [Terminal-Bench-Science 0.1](https://benchaudit.org/benchmarks/terminal-bench-science-0-1/) | Public / Aging | Math & Scientific Reasoning | #28/41 | 37.2p |
| [MMMU-Pro](https://benchaudit.org/benchmarks/mmmu-pro/) | Public / Aging | General Knowledge & QA | #14/22 | 35.5p |
| [LiveBench](https://benchaudit.org/benchmarks/livebench/) | Private / Fresh | General Knowledge & QA | #24/28 | 30.7p |
| [SWE Atlas - Codebase QnA](https://benchaudit.org/benchmarks/swe-atlas-qna/) | Public / Aging | Coding & SWE | #14/15 | 28.6p |
| [Code Migration v1.1](https://benchaudit.org/benchmarks/code-migration-v1-1/) | Private / Fresh | Coding & SWE | #18/24 | 27.9p |
| [ARC-AGI-2](https://benchaudit.org/benchmarks/arc-agi-2/) | Private / Fresh | General Knowledge & QA | #19/22 | 25.4p |
| [AutomationBench-AA](https://benchaudit.org/benchmarks/automationbench-aa/) | Private / Fresh | Agentic & Enterprise | #39/50 | 22.4p |
| [BioMysteryBench](https://benchaudit.org/benchmarks/biomysterybench/) | Public / Aging | Math & Scientific Reasoning | #22/23 | 20.8p |
| [GDPval-AA v2.1](https://benchaudit.org/benchmarks/gdpval-aa-v2-1/) | Public / Aging | Agentic & Enterprise | #34/43 | 18.5p |
| [AA-Briefcase v1.1](https://benchaudit.org/benchmarks/aa-briefcase-v1-1/) | Private / Fresh | Agentic & Enterprise | #41/50 | 18.4p |
| [Terminal-Bench 4.0](https://benchaudit.org/benchmarks/terminal-bench-4-0/) | Public / Aging | Coding & SWE | #41/50 | 18.4p |
| [AA-Omniscience v1.1](https://benchaudit.org/benchmarks/aa-omniscience-v1-1/) | Private / Fresh | General Knowledge & QA | #43/47 | 9.8p |
| [VIALS-AA](https://benchaudit.org/benchmarks/vials-aa/) | Semi-Exposed | Math & Scientific Reasoning | #22/23 | 4.5p |
