AA Coding Agent Index v2.0 - SWE-Atlas-QnA-AA: leaderboard
SWE-Atlas-QnA-AA component of AA's Coding Agent Index v2.0: Scale AI's 124 codebase question-answering tasks with AA-adjusted grading (127 of 1,306 rubric criteria reworded, GPT-6 Luna xhigh judge, five votes per criterion), answered by each model inside the coding-agent harness named in its row over three attempts per task. Task resolve rate (%).
Metric: Task Resolve Rate (%). Source: artificialanalysis.ai. Saturation forecast: Around 2029. 30 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Muse Wonder Full (Max) | 68.82 | #12 (Muse Wonder Full) |
| 2 | Gemini 4 Carbon (High) | 63.71 | #1 (Gemini 4 Carbon) |
| 3 | Grok 4.7 (xHigh) | 62.9 | #26 (Grok 4.7) |
| 4 | DeepSeek V4.1 Flash (1017) (Max) | 61.56 | #27 (DeepSeek V4.1 Flash (1017)) |
| 5 | Muse Wonder (Max) | 61.56 | #18 (Muse Wonder) |
| 6 | DeepSeek V4.1 Flash (Max) | 61.02 | #37 (DeepSeek V4.1 Flash) |
| 7 | Kimi K3 (Max) | 59.95 | #35 (Kimi K3) |
| 8 | GPT-6.1 Sol (Max) | 59.95 | #7 (GPT-6.1 Sol) |
| 9 | GPT-6 Astra (Max) | 59.14 | #6 (GPT-6 Astra) |
| 10 | mistral-large-4-oss (High) | 58.87 | #60 (mistral-large-4-oss) |
| 11 | GLM-5.3 (Max) | 58.6 | #44 (GLM-5.3) |
| 12 | Muse Spark 1.3 (Max) | 58.6 | #31 (Muse Spark 1.3) |
| 13 | Step 5 Preview | 58.6 | #53 |
| 14 | Qwen 3.8 Max | 56.18 | #34 |
| 15 | Gemini 4 Argon (High) | 55.11 | #9 (Gemini 4 Argon) |
No result here: #3 Claude Opus 5.5, #10 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=aa-coding-agent-index-v2-0-swe-atlas-qna-aa · How It Works · Data refreshed daily, snapshot 2026-11-05.