GDPval-AA v2.1
Professional Work
Artificial Analysis GDPval-AA v2.1, evaluating 220 professional tasks with shell and web access through Stirrup. Blind pairwise judgments produce Elo ratings anchored to DeepSeek V4.1 Flash (max) at 1600. This version is kept separate from GDPval-AA v2's earlier judging and rating scale.
Top models (higher is better)
| Model | Score |
|---|---|
| Claude Opus 5.5 | 1866 |
| Muse Wonder Full | 1857 |
| Kimi K3.1 Pro | 1846 |
| Claude Sonnet 5.5 | 1838 |
| Muse Wonder | 1770 |
| Claude Fable 5.1 | 1764 |
| DeepSeek V4.1 Flash 1017 | 1746 |
| Gemini 4 Carbon | 1733 |
| Claude Opus 5 | 1730 |
| Grok 4.7 | 1722 |
| MiMo-V2.6-Pro | 1691 |
| Muse Spark 1.3 | 1681 |
| Qwen3.8 Max (0902) | 1673 |
| GLM-5.3 | 1658 |
| Grok 4.6 | 1655 |
| Ling 3.1 Flash 1019 | 1653 |
| GLM-5.3 Flash | 1649 |
| Qwen3.8-Flash-Next | 1637 |
| Gemini 4 Argon | 1631 |
| Claude Haiku 5.5 | 1628 |
| Ling 3.1 Flash | 1627 |
| GPT-5.6 Sol | 1618 |
| Qwen3.8 2.4T A95B | 1616 |
| MiMo-V2.6-Flash | 1615 |
| Qwen3.8-Max | 1612 |
| GPT-6.1 Sol | 1608 |
| DeepSeek V4.1 Flash | 1600 |
| Step 5 Preview | 1594 |
| GPT-6 Astra | 1584 |
| DeepSeek V4 Flash Vision Exp | 1555 |
| Kimi K3 | 1543 |
| GPT-6 Sol | 1517 |
| Muse Spark 1.2 | 1481 |
| GPT-6.1 Terra | 1476 |
| Sonnet 5 | 1471 |
| GPT-5.6 Luna | 1469 |
| MiniMax M3.1 Flash | 1466 |
| Opus 4.8 | 1462 |
| GPT-5.6 Terra | 1462 |
| DeepSeek V4 Pro 0813 | 1458 |