# SWE Atlas - Codebase QnA Contamination Audit

> SWE Atlas - Codebase QnA scores 4.7/10 (Tier C, High Risk) on BenchAudit as of November 2026, reflecting publicly downloadable plaintext test files and isolated offline third-party grading.

- **Canonical URL:** https://benchaudit.org/benchmarks/swe-atlas-qna/
- **Benchmark ID:** `swe-atlas-qna`
- **Hub:** Scale AI
- **Category:** Coding
- **Domains:** Coding
- **Skills:** Coding, Agentic, Tool Use, Reasoning, Long Context
- **BenchAudit Score:** 4.7/10 (Tier C - High Risk)
- **Last Audited:** 2026-11-20 (Ref 2026-11)

## Factor Scores

| Factor | Score (/10) | Weight |
| --- | ---: | ---: |
| Privacy | 0 | 40% |
| Originality | 7.2 | 25% |
| Freshness | 7 | 20% |
| Grading | 10 | 15% |

## Primary-Source Evidence

### Privacy
> "Codebase QnA consists of 124 tasks that targets the upstream capability of Software Engineering - deep code comprehension that precedes any code change."
- Source: https://labs.scale.com/leaderboard/sweatlas-qna

### Originality
> "Professional software engineers and technical experts with significant coding and agentic experience write problem statements that require multi-step reasoning across the codebase."
- Source: https://labs.scale.com/leaderboard/sweatlas-qna

### Freshness
> "Update November 2026. SWE Atlas v1.1 is live. It combines the three category leaderboards into a single, easy-to-run index covering the same 284 tasks as v1.0."
- Source: https://github.com/scaleapi/SWE-Atlas

### Grading
> "Agents are instructed to avoid modifying source code files and to clean up any temporary scripts created. A programmatic check automatically fails any task that contains code changes."
- Source: https://labs.scale.com/leaderboard/sweatlas-qna

## Top Frontier Models on SWE Atlas - Codebase QnA

| Rank | Model | Creator | Percentile |
| ---: | --- | --- | ---: |
| #1 | [Claude Fable 5.2](https://benchaudit.org/models/claude-fable-5-2__max/) | Anthropic | 100.0p |
| #2 | [Claude Fable 5.1](https://benchaudit.org/models/claude-fable-5-1__max/) | Anthropic | 100.0p |
| #3 | [Claude Opus 5.5](https://benchaudit.org/models/claude-opus-5-5__max/) | Anthropic | 97.5p |
| #4 | [Claude Sonnet 5.5](https://benchaudit.org/models/claude-sonnet-5-5__max/) | Anthropic | 88.1p |
| #4 | [Inkling 2 Small](https://benchaudit.org/models/inkling-2-small/) | Thinking Machines | 88.1p |
| #6 | [GPT-6.1 Sol](https://benchaudit.org/models/gpt-6-1-sol__max/) | OpenAI | 78.7p |
| #7 | [GLM-5.3](https://benchaudit.org/models/glm-5-3__max/) | Z AI | 69.4p |
| #7 | [GPT-6 Astra](https://benchaudit.org/models/gpt-6-astra__max/) | OpenAI | 69.4p |
| #9 | [GPT-6.2 Astra](https://benchaudit.org/models/gpt-6-2-astra__max/) | OpenAI | 60.0p |
| #11 | [Claude Haiku 5.5](https://benchaudit.org/models/claude-haiku-5-5__max/) | Anthropic | 47.5p |
