DABstep — leaderboard
Data Agent Benchmark for multi-step reasoning by Adyen: ~450 data analysis questions across easy and hard levels requiring agents to analyze CSVs and cross-reference documents.
Metric: Hard Level Accuracy (%). Source: huggingface.co. Status: saturation imminent. 28 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Haiku 4.5 | 89.95 |
| 2 | GPT-5 | 57.67 |
| 3 | Gemini 2.5 Pro | 45.24 |
| 4 | Gemini 2.5 Pro (Preview 05-06) | 41.01 |
| 5 | Claude 3.5 Sonnet (20241022) | 28.04 |
| 6 | Claude Sonnet 4 (20250514) | 19.84 |
| 7 | DeepSeek V3 | 16.4 |
| 8 | O4 Mini | 14.55 |
| 9 | Claude 3.7 Sonnet (20250219) | 13.76 |
| 10 | O3 Mini | 13.76 |
| 11 | GPT-4.1 | 12.43 |
| 12 | O1 | 11.11 |
| 13 | Gemini 2.0 Flash | 9.79 |
| 14 | Claude 3.5 Sonnet | 9.26 |
| 15 | GPT-4.1 Mini | 8.99 |
Interactive version: theaggregate.ai/benchmark?slug=dabstep · How the rankings work · Data refreshed daily, snapshot 2026-07-22.