# SWE Atlas

Tests whether AI coding assistants can do everyday professional software work beyond fixing bugs: answering questions about a codebase, writing tests and refactoring code.

Category: coding. Type: leaderboard. Status: active.
Metric: Overall SWE Atlas v1.1 index score (%) with ± confidence margin (higher is better).
Source: Scale AI Labs — https://labs.scale.com/leaderboard/swe-atlas, results as of 2026-11, retrieved 2026-11-17, license: not stated.

| Rank | Model | Score (%) | ± margin |
|---|---|---|---|
| 1 | Fable 5.2 | 80.55 | 4.80 |
| 2 | Opus 5.5 | 77.45 | 5.05 |
| 3 | Fable 5.1 | 76.45 | 5.16 |
| 4 | Gemini 4 Carbon | 71.45 | 5.46 |
| 5 | Gemini 4 Argon | 70.08 | 5.48 |
| 6 | GPT-6 Astra | 69.42 | 5.55 |
| 7 | Sonnet 5.5 | 68.28 | 5.64 |
| 8 | GPT-6.1 Sol | 65.88 | 5.74 |
| 9 | Muse Wonder 1.0 | 64.83 | 5.77 |
| 10 | Qwen 3.8 Max | 64.40 | 5.71 |
| 11 | Grok 4.7 | 62.13 | 5.79 |
| 12 | DeepSeek V4.1 Flash | 62.12 | 5.85 |
| 13 | Inkling 2 Small | 61.89 | 5.83 |
| 14 | Kimi K3.1 Pro | 61.11 | 5.88 |
| 15 | Kimi K3 | 60.98 | 5.89 |
| 16 | Haiku 5.5 | 59.43 | 5.92 |
| 17 | GLM 5.3 | 58.10 | 5.90 |
| 18 | GPT-6.1 Terra | 57.69 | 5.94 |
| 19 | Gemini 3.8 Flash | 57.33 | 5.83 |
| 20 | Muse Ember 1.0 | 55.99 | 5.98 |
| 21 | GLM 5.3 Flash | 55.33 | 5.94 |
| 22 | GPT-6 Luna | 50.80 | 6.02 |
| 23 | Mistral Large 4 | 45.19 | 5.83 |
| 24 | Muse Spark Open | 42.28 | 5.90 |
| 25 | MiniMax M3 | 39.26 | 5.75 |

Notes: SWE Atlas v1.1 (Nov 2026), 284 tasks; not comparable to v1.0. Judge: GPT-6.1 Sol majority of 3.
