# Scale Labs SWE Atlas - QnA (SWE Atlas - QnA)

> A Scale Labs public leaderboard mirrored as display-only reference data. It does not affect BenchLM rankings.

Canonical page: https://benchlm.ai/benchmarks/scale-swe-atlas-qna

- Category: [Coding](/coding)
- Last updated: November 12, 2026 snapshot

## About SWE Atlas - QnA

- Year: 2026
- Tasks: 28 published rows
- Format: Published Scale leaderboard score
- Difficulty: External agent and model evaluation
- Paper: [Scale Labs leaderboard](https://labs.scale.com/leaderboard/swe-atlas)

BenchLM mirrors 28 published rows from the SWE Atlas - QnA public table captured on November 12, 2026 snapshot.

SWE Atlas - QnA is currently displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.

## Leaderboard (28 models)

| Rank | Model | Configuration | Creator | Score |
|------|-------|---------------|---------|-------|
| 1 | [Fable 5.2](https://labs.scale.com/leaderboard/swe-atlas) | SWE Atlas - QnA | Anthropic | 83.90% |
| 2 | [Qwen 3.8 Max](https://labs.scale.com/leaderboard/swe-atlas) | SWE Atlas - QnA | Alibaba | 81.50% |
| 3 | [Fable 5.1](https://labs.scale.com/leaderboard/swe-atlas) | SWE Atlas - QnA | Anthropic | 79.00% |
| 4 | [Opus 5.5](https://labs.scale.com/leaderboard/swe-atlas) | SWE Atlas - QnA | Anthropic | 77.40% |
| 5 | [Inkling 2 Small](/models/inkling-2-small) | SWE Atlas - QnA | Thinking Machines Lab | 74.20% |
| 6 | [Sonnet 5.5](https://labs.scale.com/leaderboard/swe-atlas) | SWE Atlas - QnA | Anthropic | 74.20% |
| 7 | [Gemini 4 Carbon](/models/gemini-4-carbon) | SWE Atlas - QnA | Google | 73.40% |
| 8 | [Grok 4.7](/models/grok-4-7) | SWE Atlas - QnA | xAI | 73.40% |
| 9 | [Muse Wonder 1.0](https://labs.scale.com/leaderboard/swe-atlas) | SWE Atlas - QnA | Meta | 73.40% |
| 10 | [DeepSeek V4.1 Flash 1017](/models/deepseek-v4-1-flash) | SWE Atlas - QnA | DeepSeek | 71.00% |
| 11 | [GPT-6.1 Sol](/models/gpt-6-1-sol) | SWE Atlas - QnA | OpenAI | 71.00% |
| 12 | [GLM-5.3](/models/glm-5-3) | SWE Atlas - QnA | Z.AI | 70.20% |
| 13 | [GPT-6 Astra](/models/gpt-6-astra) | SWE Atlas - QnA | OpenAI | 70.20% |
| 14 | [Kimi K3.1 Pro](/models/kimi-3-1-pro) | SWE Atlas - QnA | Moonshot AI | 70.20% |
| 15 | [GLM-5.3-Flash](/models/glm-5-3-flash) | SWE Atlas - QnA | Z.AI | 66.90% |
| 16 | [Gemini 4 Argon](/models/gemini-4-argon) | SWE Atlas - QnA | Google | 65.30% |
| 17 | [Kimi K3](/models/kimi-k3) | SWE Atlas - QnA | Moonshot AI | 65.30% |
| 18 | [Haiku 5.5](https://labs.scale.com/leaderboard/swe-atlas) | SWE Atlas - QnA | Anthropic | 62.10% |
| 19 | [Mistral Large 4](/models/mistral-large-4) | SWE Atlas - QnA | Mistral | 61.30% |
| 20 | [Muse Ember 1.0](https://labs.scale.com/leaderboard/swe-atlas) | SWE Atlas - QnA | Meta | 61.30% |
| 21 | [GPT-6.1 Terra](/models/gpt-6-1-terra) | SWE Atlas - QnA | OpenAI | 54.00% |
| 22 | [GPT-6 Luna](/models/gpt-6-luna) | SWE Atlas - QnA | OpenAI | 50.80% |
| 23 | [Muse Spark Open](/models/muse-spark-open) | SWE Atlas - QnA | Meta | 50.00% |
| 24 | [MiniMax M3](/models/minimax-m3) | SWE Atlas - QnA | MiniMax | 49.20% |
| 25 | [Gemini 3.8 Flash](/models/gemini-3-8-flash) | SWE Atlas - QnA | Google | 45.20% |
| 26 | [Nemotron 3.5 Super VL](https://labs.scale.com/leaderboard/swe-atlas) | SWE Atlas - QnA | Nvidia | 26.60% |
| 27 | [Nemotron 3 Ultra](/models/nemotron-3-ultra) | SWE Atlas - QnA | NVIDIA | 21.80% |
| 28 | [GPT-OSS 120B](/models/gpt-oss-120b) | SWE Atlas - QnA | OpenAI | 10.50% |

## FAQ

### What does SWE Atlas - QnA measure?

A Scale Labs public leaderboard mirrored as display-only reference data. It does not affect BenchLM rankings.

### Which model leads the published SWE Atlas - QnA snapshot?

Fable 5.2 currently leads the published SWE Atlas - QnA snapshot with a score of 83.90%.

### How many models are evaluated on SWE Atlas - QnA?

The November 12, 2026 snapshot contains 28 AI models.

### Does SWE Atlas - QnA affect BenchLM's overall score?

Not directly. SWE Atlas - QnA is still displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.