# GDPval-AA

Tests AI on real office-style professional tasks across many occupations, judged head-to-head.

Category: agents. Type: leaderboard. Status: active.
Metric: Elo rating from pairwise comparisons (higher is better) (higher is better).
Source: Artificial Analysis — https://artificialanalysis.ai/evaluations/gdpval-aa, retrieved 2026-11-17, license: not stated (site has Terms of Use).

| Rank | Model | Elo | Release date |
|---|---|---|---|
| 1 | Claude Opus 5.5 (Max, Default Fallback) | 1866 | Sep 2026 |
| 2 | Kimi K3.1 Pro | 1861 | Oct 2026 |
| 3 | Muse Wonder Full (Max) | 1857 | Oct 2026 |
| 4 | Claude Sonnet 5.5 (Max, Default Fallback) | 1838 | Sep 2026 |
| 5 | Claude Opus 5.5 (Xhigh, Default Fallback) | 1836 | Sep 2026 |
| 6 | Claude Fable 5.2 (Max, Default Fallback) | 1831 | Nov 2026 |
| 7 | Qwen4 Max | 1809 | Nov 2026 |
| 8 | Muse Wonder (Max) | 1770 | Oct 2026 |
| 9 | Claude Fable 5.1 (Max, Default Fallback) | 1765 | Sep 2026 |
| 10 | Claude Fable 5.2 (Xhigh, Default Fallback) | 1751 | Nov 2026 |
| 11 | DeepSeek V4.1 Flash 1017 (Max) | 1746 | Oct 2026 |
| 12 | Claude Sonnet 5.5 (Xhigh, Default Fallback) | 1743 | Sep 2026 |
| 13 | Claude Fable 5.1 (Xhigh, Default Fallback) | 1741 | Sep 2026 |
| 14 | Gemini 4 Carbon (High) | 1733 | Oct 2026 |
| 15 | Claude Opus 5 (Max) | 1729 | Jul 2026 |
| 16 | Claude Fable 5.2 (High, Default Fallback) | 1718 | Nov 2026 |
| 17 | Grok 4.7 (High) | 1718 | Sep 2026 |
| 18 | Grok 4.7 (Xhigh) | 1717 | Sep 2026 |
| 19 | Claude Opus 5.5 (High, Default Fallback) | 1709 | Sep 2026 |
| 20 | Claude Opus 5 (Xhigh) | 1695 | Jul 2026 |
| 21 | MiMo-V2.6-Pro | 1692 | Sep 2026 |
| 22 | Muse Spark 1.3 (Max) | 1681 | Sep 2026 |
| 23 | Qwen3.8 Max (0902) | 1674 | Sep 2026 |
| 24 | Grok 4.6 (Xhigh) | 1658 | Aug 2026 |
| 25 | Ling 3.1 Flash 1019 | 1655 | Oct 2026 |

Notes: GDPval-AA v2.1, top 25 of 319 models. Site terms of use not reviewed in depth.
