# ARC-AGI-2 Contamination Audit

> ARC-AGI-2 scores 8.2/10 (Tier A, Low Risk) on BenchAudit as of November 2026, reflecting a hybrid public and private held-out split and isolated offline third-party grading.

- **Canonical URL:** https://benchaudit.org/benchmarks/arc-agi-2/
- **Benchmark ID:** `arc-agi-2`
- **Hub:** ARC Prize Foundation
- **Category:** General
- **Domains:** Math, Science
- **Skills:** Reasoning, Multimodal
- **BenchAudit Score:** 8.2/10 (Tier A - Low Risk)
- **Last Audited:** 2026-11-17 (Ref 2026-11)

## Factor Scores

| Factor | Score (/10) | Weight |
| --- | ---: | ---: |
| Privacy | 7.5 | 40% |
| Originality | 10 | 25% |
| Freshness | 6 | 20% |
| Grading | 10 | 15% |

## Primary-Source Evidence

### Privacy
> "Public Eval Set 120 tasks Calibrated, public, all tasks solved pass@2 by at least two humans, use to test your systems. Semi-Private Eval Set 120 tasks Calibrated, not public, all tasks solved pass@2 by at least two humans, used for Kaggle live contest leaderboard and ARC Prize leaderboard."
- Source: https://arcprize.org/arc-agi/2

### Originality
> "The Abstraction and Reasoning Corpus for Artificial General Intelligence (ARC-AGI), introduced in 2019, established a challenging benchmark for evaluating the general fluid intelligence of artificial systems via a set of unique, novel tasks only requiring minimal prior knowledge."
- Source: https://arxiv.org/abs/2505.11831

### Freshness
> "By Greg Kamradt Published 24 Mar 2025 ARC-AGI-2 + ARC Prize 2025 is Live!"
- Source: https://arcprize.org/blog/announcing-arc-agi-2-and-arc-prize-2025

### Grading
> "Internet access is not available during Kaggle evaluation (no API-based systems like GPT/Claude/etc.)"
- Source: https://arcprize.org/competitions/2026

## Top Frontier Models on ARC-AGI-2

| Rank | Model | Creator | Percentile |
| ---: | --- | --- | ---: |
| #1 | [GPT-6.2 Astra](https://benchaudit.org/models/gpt-6-2-astra__max/) | OpenAI | 100.0p |
| #2 | [GPT-6 Astra](https://benchaudit.org/models/gpt-6-astra__max/) | OpenAI | 100.0p |
| #3 | [GPT-6.1 Sol](https://benchaudit.org/models/gpt-6-1-sol__max/) | OpenAI | 98.9p |
| #4 | [Claude Fable 5.2](https://benchaudit.org/models/claude-fable-5-2__max/) | Anthropic | 91.9p |
| #4 | [Gemini 4 Hydrogen](https://benchaudit.org/models/gemini-4-hydrogen__xhigh/) | Google | 91.9p |
| #6 | [Gemini 4 Carbon](https://benchaudit.org/models/gemini-4-carbon__high/) | Google | 85.1p |
| #7 | [Claude Opus 5.5](https://benchaudit.org/models/claude-opus-5-5__max/) | Anthropic | 80.5p |
| #8 | [Gemini 4 Argon](https://benchaudit.org/models/gemini-4-argon__high/) | Google | 75.9p |
| #9 | [Claude Fable 5.1](https://benchaudit.org/models/claude-fable-5-1__max/) | Anthropic | 71.3p |
| #10 | [GPT-6 Sol](https://benchaudit.org/models/gpt-6-sol__max/) | OpenAI | 66.7p |
