Abstract This cross-sectional study compared three LLMs (Grok 4.1, Gemini 3, and ChatGPT 5.2) in classifying ultra-processed foods (UPF) using best-selling products from leading supermarket chains covering 53.2% of the national market. Of 3001 products, 2920 with complete ingredient information were included; two trained dietitians assigned NOVA groups as the reference standard. In the reference classification, 74.3% of products were UPF. Under the baseline prompt, all models underestimated UPF prevalence compared with the reference standard (p 0.001). ChatGPT 5.2 yielded the highest binary UPF detection performance (accuracy: 69.01%; sensitivity: 59.01%; specificity: 98.00%; F1: 73.88%). Prompt sensitivity analyses revealed that a minimal prompt substantially outperformed the detailed baseline for most models (Gemini 3 F1: 94.20%; ChatGPT 5.2 F1: 92.62%). Low inter-run agreement (κ: 0.01–0.27) indicated sensitivity to model updates, supporting prompt calibration and human oversight. These findings suggest that off-the-shelf LLMs require prompt calibration and human oversight before UPF surveillance workflows.
Bayram et al. (Wed,) studied this question.