PrepBench: leaderboard
Data preparation is a central and time-consuming stage in data analysis workflows. Traditionally, commercial tools have relied on graphical user interfaces (GUIs) to simplify data preparation, allowing users to define transformations through visual operators and workflows. Recent advances in large language models (LLMs) raise the possibility of a paradigm shift toward natural language (NL)-driven
Metric: End-to-End Prep-Code Acc. (self-reported). Source: benchmarklist.com. Saturation forecast: Around March 2027. 10 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.1 Codex | 54.9 | #159 |
| 2 | Gemini 3 Flash | 53.3 | #99 |
| 3 | Claude Sonnet 4.5 | 52 | #142 |
| 4 | Kimi K2 (Thinking) | 49.7 | #243 (Kimi K2) |
| 5 | DeepSeek V3.2 | 44.8 | #205 |
| 6 | GLM-4.7 | 41.5 | #192 |
| 7 | Qwen 3 235B A22B | 38.2 | #313 |
| 8 | Devstral 2 | 33.3 | #411 |
| 9 | Grok Code Fast 1 | 30.1 | #337 |
| 10 | GPT-4o | 16.7 | #343 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=prepbench · How It Works · Data refreshed daily, snapshot 2026-10-20.