CITBench - Multi-Step: leaderboard
CITBench (2026) evaluates LLMs on tabular data processing tasks of matching, cleaning, augmentation and transformation. Scored in the offline single-turn setting as average performance: the mean Level-Accuracy over repeated runs, with partial credit, in percent; higher is better. This board covers the multi-step tier of chained operation pipelines.
Metric: Average performance (%; mean Level-Accuracy with partial credit, each of repeated runs scored 0 to 1, on the offline single-turn tabular data processing tasks of CITBench, a balanced 50% slice of its 1,296 instances in 18 matching, cleaning, augmentation and transformation task types; temperature 1.0, 8,192 output tokens; Multi-Step tier: chained operation pipelines). Source: arxiv.org. Saturation forecast: Around February 2027. 12 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Sonnet 4.5 | 43.52 | #244 |
| 2 | DeepSeek V3.2 | 41.24 | #403 |
| 3 | Gemini 3 Flash | 40.39 | #159 |
| 4 | Gemini 3 Pro | 38.13 | #135 |
| 5 | GPT-5.2 | 37.05 | #172 |
| 6 | Claude Haiku 4.5 | 34.71 | #568 |
| 7 | GLM-4.7 | 27.06 | #365 |
| 8 | Qwen 3 Coder Flash | 17.9 | #1624 |
| 9 | Qwen 3 235B A22B | 17.24 | #622 |
| 10 | Qwen 3 32B | 15.57 | #960 |
| 11 | GPT-4o | 12.93 | #741 |
| 12 | Qwen 3 14B | 4.45 | #1235 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=citbench-multi-step · How It Works · Data refreshed daily, snapshot 2026-11-12.