# Is It Still Worth Training a Classical Model in the Era of LLMs? A Crossover Benchmark on Tabular Data

arXiv: https://arxiv.org/abs/2609.20218 · HTML: https://mlsanity.com/paper/2609.20218/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 15 of its 28 references that are in the map.
- First posted on arXiv 28 Jul 2026.

## Builds on (most cited first, 15 of 15)

- [Scaling Laws for Neural Language Models](https://mlsanity.com/paper/2001.08361/) (2020)
- [XGBoost: A Scalable Tree Boosting System](https://mlsanity.com/paper/1603.02754/) (2016)
- [Extracting Training Data from Large Language Models](https://mlsanity.com/paper/2012.07805/) (2020)
- [Revisiting Deep Learning Models for Tabular Data](https://mlsanity.com/paper/2106.11959/) (2021)
- [TabPFN: A Transformer That Solves Small Tabular Classification Problems in a Second](https://mlsanity.com/paper/2207.01848/) (2022)
- [TabNet: Attentive Interpretable Tabular Learning](https://mlsanity.com/paper/1908.07442/) (2019)
- [Deep Learning Scaling is Predictable, Empirically](https://mlsanity.com/paper/1712.00409/) (2017)
- [Deep Neural Networks and Tabular Data: A Survey](https://mlsanity.com/paper/2110.01889/) (2021)
- [AutoGluon-Tabular: Robust and Accurate AutoML for Structured Data](https://mlsanity.com/paper/2003.06505/) (2020)
- [Data Contamination: From Memorization to Exploitation](https://mlsanity.com/paper/2203.08242/) (2022)
- [When Do Neural Nets Outperform Boosted Trees on Tabular Data?](https://mlsanity.com/paper/2305.02997/) (2023)
- [Neural Oblivious Decision Ensembles for Deep Learning on Tabular Data](https://mlsanity.com/paper/1909.06312/) (2019)
- [Well-tuned Simple Nets Excel on Tabular Datasets](https://mlsanity.com/paper/2106.11189/) (2021)
- [LIFT: Language-Interfaced Fine-Tuning for Non-Language Machine Learning Tasks](https://mlsanity.com/paper/2206.06565/) (2022)
- [Large Scale Transfer Learning for Tabular Data via Language Modeling](https://mlsanity.com/paper/2406.12031/) (2024)

## Abstract (arXiv)

Large language models can label a tabular row from a plain-English description with no training - a capability now shipping in mainstream spreadsheet tools such as Microsoft Copilot in Excel and Anthropic's Claude for Excel - raising a practical question for the many business prediction problems where labels are expensive: should you prompt a frozen LLM, or collect data and train a model - and if so, how much data? We quantify the answer with the labeled-data crossover N*, the training-set size at which a trained classical model's learning curve overtakes a frozen LLM's training-free (and therefore flat) error. Aggregating 126 independent student evaluations of small GPT models under eight prompting configurations across 18 tabular datasets, paired with authoritative power-law learning curves for six classical model families, we find that training wins fast: even given an oracle choice of its best prompt configuration, a trained classical model beats the small frozen LLM using no more labeled data than is already on hand in 86% of cases, and wins by the smallest labeled subset we evaluate in 40%, with the observed crossover at a median of ~6% of the training set. In-context few-shot examples do not behave like training - error versus shot count does not follow a power law - and the same protocol re-run by independent implementers varies with a coefficient of variation of 0.148. A controlled probe indicates the LLM depends on recognizable feature-name semantics, which plausibly makes our crossover a conservative estimate (we do not claim memorization). For a typical business table, the evidence is clear: collect a few hundred labels and train a gradient-boosted model.