VIDRAFT has launched FINAL-Bench, a new daily leaderboard that compares AI and traditional numerical weather prediction (NWP) models. The benchmark evaluates key variables like temperature and precipitation across 1- to 7-day lead times for 24 cities globally. It uses a common-case methodology to ensure fair comparisons by only scoring data points where all models have provided forecasts, and it uses ERA5 reanalysis data as its verification baseline. AI
IMPACT Provides a standardized, continuously updated benchmark for evaluating AI's performance in weather forecasting against traditional methods.
RANK_REASON Launch of a new benchmark/leaderboard tool for evaluating AI vs traditional models.
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →