Plotly benchmarks 57 LLMs on DataAnalyticsBench

Plotly evaluated 57 language models released between November 2024 and November 2026 using its DataAnalyticsBench exam to measure progress in LLM data analysis capabilities. The analysis found that top-performing models progressed from solving 20 out of 43 questions to passing all 43, while execution costs dropped significantly over the two-year period. Open article →