CDR-Bench - Order-Agnostic Recipes: leaderboard
Metric: Recipe success RS@3 (%) on the order-agnostic mapper-recipe tasks of CDR-Bench (3,462 data-refinement tasks over web, LaTeX, RAG-preparation and privacy text with 29 Data-Juicer operators and deterministic reference outputs): a task counts when the model's refined text or KEEP/DROP judgment exactly matches the reference within 3 attempts; non-thinking mode, baseline prompt; higher is better. Source: arxiv.org. Saturation forecast: Around 2032. 11 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemma 4 31B (IT) (Non-reasoning) | 25.55 |
| 2 | GLM-5 (Non-reasoning) | 24.8 |
| 3 | Kimi K2.6 (Non-reasoning) | 24.14 |
| 4 | DeepSeek V4 Flash (Non-reasoning) | 23.94 |
| 5 | Qwen 3.6 Max | 23.94 |
| 6 | Qwen 3.6 27B (Non-reasoning) | 23.73 |
| 7 | GPT-5.4 (Non-reasoning) | 22.74 |
| 8 | Claude Opus 4.6 | 22.33 |
| 9 | Claude Opus 4.5 | 22.13 |
| 10 | Qwen 3.6 35B A3B (Non-reasoning) | 20.4 |
| 11 | Llama 4 Scout | 13.59 |
Interactive version: theaggregate.ai/benchmark?slug=cdr-bench-order-agnostic-recipes · How It Works · Data refreshed daily, snapshot 2026-09-29.