LLMTabBench: leaderboard

Metric: ROC-AUC (0-1; no demonstrations; all 32 real datasets, 24 classic and 8 published after October 2024; generation mode, prompt with column names and task description; mean over datasets). Source: arxiv.org. Saturation forecast: Estimated already saturated. 4 models tracked.

Top models

#ModelScore
1Qwen 3 14B0.71
2GPT-4o Mini0.71
3Qwen 3 8B0.7
4Qwen 3 1.7B0.63

Interactive version: theaggregate.ai/benchmark?slug=llmtabbench · How It Works · Data refreshed daily, snapshot 2026-09-26.