GDPval-MM: leaderboard
Multimodal variant of GDPval evaluating models on economically valuable tasks across industries requiring processing and generating documents, slides, diagrams, spreadsheets, images, and other professional deliverables.
Metric: Wins + ties (self-reported). Source: benchmarklist.com. Status: saturation imminent. 11 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.2 | 70.9 |
| 2 | Claude Opus 4.5 | 59.6 |
| 3 | Gemini 3 Pro | 53.5 |
| 4 | Claude Sonnet 4.5 | 50.3 |
| 5 | Claude Opus 4.1 | 47.6 |
| 6 | GPT-5 | 38 |
| 7 | O3 | 34.1 |
| 8 | O4 Mini (High) | 27.8 |
| 9 | Gemini 2.5 Pro | 25.5 |
| 10 | Grok 4 | 24.3 |
| 11 | GPT-4o | 12.3 |
Interactive version: theaggregate.ai/benchmark?slug=gdpval-mm · How It Works · Data refreshed daily, snapshot 2026-09-05.